.ii'hr1GllSflOl:l~""COllllflfllllcaf
Federal Office of Road Safety
SPEED PERCEPTION 1:
DRIVERS' JUDGEMENTS
OF SAFETY AND SPEED
ON URBAN AND RURAL STRAIGHT ROADS
,Prepared by
RACV Limited
I\t4ELBOURNE,
AUSTRALIA
CR54
May 1987
~.
I'
'~l'i.\
DEPARTMENT OF TRANSPORT
FEDERAL OFFICE OF ROAD SAFETY
DOCUMENT
Report
CR 54
Title
RETRIEVAL
INFORMATION
No.
and
I Date
1987
I
171
o 642
I ISBN
Pages
51035
I CR=0810-770X
ISSN
Subtitle
SPEED PERCEPTION 1, Drivers' j~dgements
on urban and rural straight roads.
of safety and speed
Author (51
FILDES,
Performing
B.N., FLETCHER,
M.R. and CORRIGAN,
J.McM.
Organisation
Royal Automobile Club of Victoria (RACV) Limited,
550 Princes Highway,
NOBLE PARK, VICTORIA, 3174, AUSTRALIA.
Sponscr
Federal Office of Road Safety,
G.P.O. Box 594,
CANBERRA, A.C.T., 2601, AUSTRALIA.
Ava.ilable
from
Federal
Office
of Road Safety
-iPrice/AvailabilitY/Pormat
Abstract
This project was undertaken to evaluate the role of the type of road and
road width, roadside development, travel speed, driving experience and
the sex of the driver on estimates, of safety and travel speed. A
preliminary study showed that laboratory testing using 16mm movie film
stimulus materials was suitable for testing road speed perception.
Three factorial design experiments were performed using urban, rural
and semi-rural road scenes in which the independent variables ,were
systematically manipulated. Slash-line responses of safe operating
speed and travel speed est~tes
in km/h were'collected from 108
licensed. drivers, recruited as subjects. Free speed measurements were
also collected at each test site where practical. The results showed
that the presentation speed had the strongest effect on. the subjects'
judgements. The type of road also influenced speed perception while
roadside development effects were dependent upon the type,of environment
(urban or rural) and the 'road category. Driver experience did not unduly
affect the results, although'the sex of the driver had a slight
influence on the data. Recommendations are made for further research
in this area to establish potential countermeasures against speeding.
Keywords
SPEED, PERCEPTION,
WOMAN,
HIGHWAY, CARRIAGEWAY,
PAVEMENT,
EXPERIENCE (HUMAN), RECENTLY QUALIFIED
ENVIRONMENT,
DRIVER,
LABORATORY,
MAN,
SAFETY
NOTES.
(1) FORS reports
are disseminated
in the interest
of infomation
exchanq8.
(2) The views exprea~ed
are those of the author(s)
and do not necessarily
represent
those of the Commonwealth
Government.
()) The Federal
Office
of Road Safety
publiShes
two series
of reports
(a) reports
qenerated
as a result of research
done within
FORS are
published
in the OR series;
Cb) reports
of research
conducted
by other organisations
on behalf
of FORS are published
in the CR series.
-~I
i
!
TABLE OF CONTENTS
ACKNOWLEDGEMENTS
Executive
1.
The authors
are indebted
to the Federa~
Office
of Road
Safety,
Federa~
Department
Of Transport,
Austra~ia,
for their
sponsorship,
interest
and assistance
in this project,
Summary
INTRODUCTION
1.1 STUDY OBJECTIVES
1.2 STATEMENT OF THE PROBLEM
A study Of this magnitude
cou~d not have been undertaken
without
the he~p and co-operation
Of a great number
of peop~e,
J, Triggs of Monash
In particu~ar,
Associate
Professor
Thomas
university,
Me~bourne,
Austra~ia
and Dr. Burton W. Stevens,
Senior
Research
Psycho~ogist
of the Federa~
Highway
Administration,
Virginia,
USA were extreme~y
he~pfu~
throughout
the project.
The staff at RACV Limited
were a~so most
Russe~~
and Mr. John Sanderson
of the Traffic
generous~y
assisted
in the project
design
and
report,
the Chief Engineer
Mr. John McKenzie
research
vehic~e.
whi~e Mr. Ron McLennan
and
Park Headquarters
provided
the ~aboratory
and
invo~ved
in the experiment.
Mr A~an Dow~ing
Richardson
provided
a~~ the artwork.
Higher
Mr. Don Hausser
and
Education
Research
3
1.3.1 Speed and the Number of Crashes
6
1.3.2 Speed and Crash Severity
1.4 POSTED SPEED LIMITS
6
7
1.4.1 The Speed Zone Index
he~pfu~.
Mr. Ian
& Safety Department
preparation
Of the
provided
the
his staff at Nob~e
many of the peop~e
and Ms Ji~~
Preparation
of stimu~us
materia~s
was aided considerab~y
thanks
to Professor
R.H. Day and the Department
of Psycho~ogy
staff at Monash
University.
In particu~ar,
Mr. Ken Ha~~ and
Mr. V~adimar
Kohout
arranged
for the ~oan of the Bo~ex camera
Mr. Richard
HObbs,
Senior
Technica~
Officer,
constructed
the
vehic~e
mount
used on the research
vehic~e.
2
1.3 SPEED AND CRASH INVOLVEMENT
1.5 DEFINING
1.6 PREVIOUS RESEARCH IN SENSORY PERCEPTION
1.6.1 Visual Performance Studies
1.6.2 Absolu~e Judgements O~ Speed
1.6.3 Retinal Streaming
2. POSSIBLE CUES FOR SPEED PERCEPTION
2.1 DATA SOURCES
2.2 URBAN AND RURAL ENVIRONMENTS
2.3 ROAD ALIGNMENT
and
Mr. John Torrance
of Monash university's
Unit kind~y
he~ped
with fi~m editing.
9
10
10
11
11
13
14
14
16
2.4 ROAD CATEGORY AND LANE WIDTH
18
2.5 ROAD MARKINGS AND DELINEATION
20
2.6 SIGHT DISTANCE AND GRADIENT
21
2.7 OTHER TRAFFIC AND DENSITY
2.8 NIGHT AND DAY
22
2.9 ROADSIDE DEVELOPMENT
24
2.10 DRIVING EXPERIENCE
2.11 WEATHER
2.12 PARKED VEHICLES
2.13 SUMMARY
3.
8
SPEED PERCEPTION
AND PEDESTRIANS
25
26
27
28
THE MEASUREMENT OF SPEED AND PERCEPTION
3.1 ON-ROAD STUDIES
3.2 SPEED ASSESSMENT
23
IN THE LABOROTORY
31
31
32
3.3 STATIC AND DYNAMIC DISPLAYS
33
3.4 FREE SPEED DATA ON THE ROAD
34
3.5 SPEED MEASUREMENT
35
~------------~
IN THE LABORATORY
IS
~..~~"
,"n'''_
~~'~'H'_-'
~_"~·~O_<·
__ .•• '
'·_u
- » .•• -- .•
3.5.1 Physical Measures of Speed
3.5.2 Numerical Measures for Speed
3.5.3 Subjective
3.6 SUMMARY
4.
RESEARCH
Magnitude
Scaling
4.2 STIMULUS
4.3 FREE SPEED VEHICLE
DATA
36
39
40
41
FACTORS
42
SPEED
4,7 DETAILED EXPERIMENTAL DESIGN
SUBJECTS
DATA
ANALYSIS
STIMULUS
MATERIALS
LABORATORY
VALIDATION
4.8
STIMULUS
MATERIALS STUDY
42
44
8.
49
49
57
57
62
5.5 DISCUSSION
6.
55
60
THE RURAL EXPERIMENT
65
6.1 STIMULUS MATERIALS
65
6.2 EXPERIMENTAL
67
PROCEDURE
68
6.3 RESULTS
6,3.1 Safe Operating
Speed Responses
6.3.2 Speed Estimation
Errors
6.3.3 Free Speed Measurements
6.4 DISCUSSION
6.4.1 Presentation
Speed
6.4.2 Type of Road
6.4.3 Roadside Environment
6.4.4 Driver Variables
68
70
74
83
83
87
7,3.3 Free Speed Measurements
7.4 DISCUSSION
92
7.4.1 Presentation Speed
7.4,2 Type of Road
7.4.3 Roadside Environment
94
94
95
96
97
THE URBAN EXPERIMENT
99
99
PROCEDURE
101
101
101
104
8.3.3 Free Speed Measurements
8.4 DISCUSSION
52
54
5.4.2 Speed Estimation Data
5,4,3 Free Speed Data
5.4.4 Road & Environment Effects
83
PROCEDURE
8.3.1 Safe Operating Speed Responses
8.3.2 Speed Estimation Errors
49
Speed Data
7.2 EXPERIMENTAL
7.3 RESULTS
8.2 EXPERIMENTAL
8.3 RESULTS
46
5.4.1 Safe Operating
81
81
8.1 STIMULUS MATERIALS
46
5.4 RESULTS
EXPERIMENT
7.4.4 Driver Variables
44
4,10
5.2 5.
ROAD
5.3
TRIALS VARIABLE
TRIALS
4.9
1
4DEPENDENT
,11LABORATORY
THE SEMI-RURAL
7.1 STIMULUS MATERIALS
7.3.1 Safe Operating Speed Responses
7.3.2 Speed Estimation Errors
40
VARIABLES
4.6 PRESENTATION
7.
35
39
PRESENTATION
4.5' CONTROLLED
-,
35
39
EXPERIMENTATION
4.4 INDEPENDENT
•
37
STRATEGY
4.1 LABORATORY
,
108
110
8.4.1 Presentation Speed
8.4.2 Type of Road
110
8.4.3 Roadside
113
111
Environment
8.4.4 Driver Variables
9.
GENERAL DISCUSSION
113
AND FINDNGS
9,1 THE RELEVANCE OF LABORATORY
9.2 PRESENTATION SPEED
115
TESTING
9.2.1 The Success of the Experimental
115
115
Method
9.2.2 Road Safety Consequences
9.3 TYPE OF ROAD
9.3.1 The Importance of the Road Pavement
9.3.2 The Role of Lane Width
9.4 ROADSIDE
ENVIRONMENT
78
78
79
--------~
9.4.1 The Discriminatory
9.4.2 Rural Environments
116
117
117
118
120
77
77
116
Effect of the Roadside
120
120
9.4.3 Semi-rural Environments
121
9.4.4 Urban Environments
121
9,5 DRIVER VARIABLES
122
i."'- ,...-~.9.6
9.5.1
Driver
9.5.2
The
SAFETY
9.6.1
9.6.2
9.7
9.9
AND
The
The
of
the
SPEED
0- "--'
-,
122
124
MEASURES
of
Role
123
Driver
Operating
AND
Speed
-.,
Experience
Safe
LIMIT
FURTHER
9.9.1
•.
Estimate
SPEED
9.8.1
•• -
Sex
PERCEPTION
9.7.1
9,8
"0
Travel
FREE
of
Speed
Response
ON
Perception
THE
125
ROAD
Additional
127
IN SPEED
Variables
9.9.2
Follow-up
9.9.3
Countermeasures
9.9.4
Speed
Anomalous
9.9.5
Perception
Zone
for
Testing
Findings
Excessive
Speed
Other
Driving
REFERENCES
A - Details
ATTACHMENT
B - Experimental
of
ATTACliMENT
C - ANOVA
Figure
4.0
Test
by
travel
Research
with
speed,
4
Sites.
Manoeuvres
.••
camera
in
45
128
4.l
129
Figure
5.0
Laboratory
validation
Figure
5.l
Pilot
sites
Sample
page
from
the
response
booklet
sites
road
Metropolitan
131
133
Figure
5.2
150
Figure
5.3
Details
Safe
165
51
laboratory
operating
obtained
(Melbourne
Area)
of
53
arrangement
speed
in the
responses
laboratory
validation
study
Figure
Figure
5.4
5.5
56
Errors
in travel
from
the
Free
speed
at
8 of
5.6
Safe
of
6.0
Rural
Figure
6.1
Safe
for
Figure
6.2
of
6.3
in
the
59
study
speed
judgement
studied
validation
58
in kro/h taken
sites
factors
for
in
a
the
61
study
66
operating
speed
roads
speed
interest
Travel
study
sites
experiment
Figure
12 test
road
rural
Travel
validation
validation
laboratory
Figure
estimates
measurements
the
operating
range
speed
laboratory
laboratory
Figure
47
50
129
Instructions.
Tables
vehicle
position
162
Summary
Involvement
rate
day and night
.
Figure
130
Index
ATTACHMENT
1.1
127
PERCEPTION
Against
and
Figure
126
Index
RESEARCH
Page
125
in Driving
DETERMINATION
Zone
OF FIGURES
124
Speed
SPEED
LIST
124
results
of
errors
for
for
rural
the
roads
interest
1
in experiment
main
69
effects
in
1
speed
interactions
in experiment
71
errors
observed
1.
for
the
for
significant
rural
roads
72
Figure 6.4
Free speed variations in kro/h
for type of road and roadside
Figure
7.0
Figure
7.1
Figure
7.2
Figure
7.3
Figure
7.4
Semi-rural
effects of interest
roads in experiment
for semi-rural
2
Travel
for the
speed errors
roads in experiment
7.6
84
of
85
88
2 .
2
89
90
Free speed variations in km/h
for type of rOud and roadside
environment for semi-rural
in experiment 2
Figure
8.0
Urban
Figure
8.1
Safe operating
Figure
8.2
Travel
roads
93
sites
100
speed results
for urban roads in experiment
speed errors
effects of interest
in experiment 3
of interest
3
102
for the main
for urban roads
--------j
observed
8.4
Free speed variations
for
3 .
106
in km/h for
for urban roads in experiment
Travel speed errors for the significant
4- way interaction for semi-rural
roads in experiment
interaction
type of road and roadside environment
2- and 3- way interactions
for semi-rural
Figure
Figure
82
speed errors of the main
significant
Figure 7.5
76
Safe operating speed main effects of
interest for semi-rural roads in
experiment 2
Travel
speed errors for the
urban roads in experiment
sites
Safe operating speed interactions
interest for semi-rural roads in
experiment 2
Travel
significant
environment for rural roads in
experiment 1 .
Figure
8.3
105
i
3 ..
109
EXECUTIVE
trsT OF TABLES
SUMMARY
Page
Table 1.1
Table 2.1
Table 2.2
Relationship between
involvement .
Table 4.1
II
13
of crash causation
studies which implicate
geometry .
Table 2.3
limits
6
Possible road and environment
variables
Summary
Design
speed and crash
in
roadside
features
There
features
road
rates
Summary of the effects reported of
the road and environment factors
reviewed
is
and
_17
how
29
information
design
these
and
engineers
that
speed
affect
perception
on
research
their
These
not
different
by drivers
about
travel
automatic
(less
conscious)
that
the perceptual
influence
of
include
much
speed
road
crash
is known
on
the
about
road_
of human
levels
higher-order
speed.
and
road.
can
perception
speed
characteristics
evidence
However.
involves
processing.
determining
road
the
surrounds
the
for
various
performance.
features
relatively
43
by
considerable
vehicle
Speed
for the
used
assume
influence
of roads
decisions
Detailed experimental
main experiments
criteria
Australia
as well
processes
cognitive
as
the
involved
in
'.",
sensory
perception.
Research
involved
,!
in
evidence
sensory
improvements
in
The
Project
Research
suggests
perception
road
that
on
manipUlating
the
road
may
the
lead
visual
cues
to long-term
behaviour.
-
l
This
evaluate
speed
.
!I
t'fo-
the
the
road
effects
Judgements
speed
environment
perception
project
sensory
A literature
,-
t:
~-
research
assessing
of
several
out
speed
road
and
review
human
of speed,
highlighted
In addition.
factors.
were also
likely
on
a suitable
the
roadside
various
a number
road
means
and
features
}~;
i
i
methods
of
to influence
identified.
~
j
to develop
of
on
of testing
road.
the
Of
to
of drivers.
perception.
and
set
perception
sensory
the
A validation
speed
study
perception
experimental
on
was
the
performed
road
CQUld
be
initially
to see
simulated
accurately
The
whether
Main
Findinqs
in an
laboratory.
The
strongest
presentations
Twelve
road
encompassed
Free
sites
a range
speed
were
selected
Of different
traffic
patterns
around
roads
were
Melbourne
and
collected
that
while
roadside
environments.
at
sites.
these
Passengers
operating
of
the
the
in
a research
assessments
12 sites.
subjects
of
in
speed
Similar
sites,
taken
were
were
laboratory
from
asked
to estimate
responses
an experimental
same
vehicle
and
the
also
using
driving
to make
travel
in
made
film
position
environment,
was
result
speed.
to be
speed
limit
were
assessed
not
sex
travel
Judged
operating
unduly
degree,
by the
for
were
a safe
was
to some
and
the
was
limit
by
of
affected
the
the
speed
type
driver
by
of
on
for
slow
as
too
area
but
the
the
road
Film
too
and
roadside
occasions.
at each
This
result
findings
segments
of
This
observed
speed
above
of what
influenced,
by other
16mm
terms
subjects.
safe
speed
the
presentations
fast
was
effect
below
any
an
and
speed
was
consistent
confirmed
perception
the
with
need
previous
for
moving
speed
stimulus
estimation
materials
in
experimentation.
automobile.
The
The
results
of
this
study
responses
in
the
laboratory
collected
on
the
road.
were
was
However,
encouraging.
almost
The
identical
safety
to
responses
pattern
of
those
in
with
,!l
I]
i
the
I
laboratory
road
were
feedback
heightened
presumably
because
of a reduction
in
type
resulted
in
estimates
function
of
may
have
research
undertaken
multi-factorial
involving
environments.
The
development,
all
travel
systematically
laboratory
rural,
role
experiments
semi-rural
of
road
speed,
driver
evaluated
on
and
category.
urban
road
experience
estimates
were
and
of
of
is
driving
and
eight
experience
were
testing
method
sex
and
The
data
collected
omega-squared
speed
greater
and
of
This
finding,
however,
greater
whether
the
was
road
Roads
generally
speed.
and
safety
perception.
width
underwas
also
a
or
not.
of
travel
divided
licensed
recruited
developed
drivers
and
in
with
tested
the
varying
using
previous
were
analysed
using
analysis
suggestion
that
speed
the
particular
perception,
to confirm
environment
under
In rural
as
those
travel
degrees
this
lane
although
further
finding.
effects
were
very
much
dependent
on
the
test.
environments,
safer,
heavily
and
roads
travel
treed.
without
speeds
This
was
roadside
trees
under-estimated
especially
so
were
mUCh
for
more,
faster
speeds.
the
For
study.
of
variance
their
the
statistics.
ii
some
influenced
were
travel
generally
and
and
estimates
region
required
environment
roadside
driver
safety
was
Roadside
road
width,
than
laboratory
the
also
perceived
hundred
travel
influenced
then
speed.
One
higher
also
standard
i
information.
Experimentation
Three
road
design
of
There
Laboratorv
of
a higher
j
semi-rural
had
effect
degree
less
was
of
environments,
dependent
urban
though.
on speed
influence
on
development
the
roadside
perception.
type
of
road
trees
Moreover,
and,
seemingly,
on
as well.
- iii -
-
Urban
roadside
environments
and
semi-rural
settings,
rural
commercial
or
perceived
industrial
to be
settings,
The
effect
decrease
the
to increase
comprising
safer
these
of
than
Of
and
safety
estimates
different
either
industrial
were
only
industrial
for
4-lane
on divided
settings
and
or
road
were
perceptions.
was
2-lane
results
but
when
roadS.
travel
and
free
to interpret
they
facilitate
in
in
compliment
a more
speed
terms
the
detailed
measures
on
the
of driver
safe
operation
account
of
the
speed
role
of
driving.
free
for
speed
difficult
However,
and
The
to
roads
Of
more
perception
small.
complexes
were
responses
commercial
very
and
Estima~es
to
residential
Residential
differences
commercial
estimates
safety
noticeab~y
developments.
slightly
a~though
were
speed
type
the mean
of
speed
data
generally
road
and
values
supported
roadside
were
the
perceptual
environment,
equal
especially
to or lower
tttan the
speed
limit.
This
service
and
resu~t
roads
was
and
industrial
explained
by
off-street
complexes
the
increased
parking,
located
often
on major
spaciousness
found
of
This
at commercial
divided
effective
arterials.
speed
Driver
year
effects
drivers
in
these
responded
suggesting
that
of driving
experience.
experiments
similarly
speed
to
perception
were
minimal.
experienced
was
not
on
Thus,
any
skill
way
differences,
these
speed
drivers
two
rather
groups
differences
more
likely
than
process
apparent
invo~ve
Research
The
this
in
sex
of
estimates
may
future
have
the
in
been
speed
driver
rural
had
a chance
perception
and
effect
at
finding.
high
difference
it
on
FOllOW-Up
in
variables
the
before
both
travel
deserves
will
road
when
be most
drivers'
safe.
Required
safety
made
for
additional
research
in road
inVOlving
other
the
anomalous
findings,
countermeasures
against
resolution
and
of
of
likely
speeding
perceptual
is necessary
can
be
tested.
and
speeds.
testing
are
experimentation,
and
a range
Further
While
perceptual
further
formulate
researCh.
estimate
Other
overly
the
novice
or driving
information.
some
areas
not
on
perception.
determined
speed
are
countermeasures
speeds
amount
between
cognitive
a fundamental
sensory
perceptual
vehicle
Recommendations
on-road
experienced
perceptions
Additional
the
speed
and
that
First
drivers,
dependent
suggests
at reducing
urban
research
is
also
factors
involved
in
the
a mathematica~
of travel
procedure
required
speed
zone
to arrive
to identify
index
at
and
all
the
to
an unbiased
speed.
Findinqs
The
assessing
consistent
interpreted
safe
operating
speed
and
speed
perception.
the
simpLy
results
and
response
The
from
data
this
meaningfUlly.
was
a suitable
Obtained
teChnique
was
means
reliable
could
of
and
I·
t·
r
be
i:
~
r:
j
,
, ..
- iv -
- v -
~
1.
INTRODUCTION
The Federal Office of Road Safety commissioned
in April 1986 to undertake
changes on drivers'
identifying
speed estimates.
how the environment
through a driver's
RACV Limited
a study of the effects of perceptual
perception
The study was aimed at
influenced
speed behaviour
of the road ahead.
The long-term aim of the study was to identify environmental
countermeasures that could be used to control speed behaviour in
addition
to education
and law enforcement.
out to further increase
driver perception
The study also set
the general understanding
in speed behaviour
of the role of
on the road.
1.1 STUDY OBJECTIVES
The three objectives
specified
Safety for the speed perception
To establish
by the Federal Office of Road
project were;-
which factors
influence
speed choice at sites
representative of a variety of situations
urban and rural areas in Australia.
in appropriate
To measure and report upon the operating speeds and other
relevant traffic characteristics at the selected sites.
To report upon the relationship,
identified
observed.
if any, between
the factors
and the speed and traffic characteristics
I··
.
_~S':~
i.J ...•
,.
The project was to be a preliminary
of perceptual
changes
on drivers'
to be placed on developing
evaluation and testing
environment factors.
speed judgements.
a suitable methodology
of the effect
Emphasis
was
for the
of some of the more important
-
"--
evaluation
road and
1 -
------------------------------------------
~;1~
1
~---
....,....
Hence, a minor literature
possible
independent
review was required to highlight
and dependent
testing was needed to validate
variables.
In addition,
the testing procedure prior to
full experimentation.
Such an approach, however, necessitates understanding the
role of environmental cues and their interactions in a driver's
speed decisions.
To date, however, very little is known about
the influence of the road and surrounding environment on a
driver's
1.2 STATEMENT
perception
of speed.
OF THE PROBLEM
1:3 SPEED AND CRASH INVOLVEMENT
The role of speed in road crashes has attracted considerable
debate over the years.
Those sharing
the enforcement
view argue that road crashes would be minimised
obeyed the present speed limits imposed
1984; Vulcan, 1986).
point of
if everyone
on our roads ( McMenomy,
Studies which evaluated the relationships
between
speed and
crash involvement and severity are limited and generally refer to
work undertaken in the 1960s. More recent studies have not
investigated
the relationships
per se but have concentrated
on
the effects of speed limit changes.
The fact is, of course, that a large proportion
consistently
1980; Cowley,
1986).
drive above the legal limits
1980; Elliot,
1981; Sanderson
Hauer, Ahlin and Bowser
of motorists
(Mostyn and Sheppard,
and Corrigan,
(1982), Armour
1984;
(1986) and Shinar
and Stiebel (1986) showed that enforcement could lead to speed
reductions when it was intermittent but then the effects tended
to be localised.
Elliot
(1981) argued that a "personal limit"
existed where drivers are controlled
by what they perceive to be
a safe speed, rather than the legal limit.
claimed that motorists
between
exceeding
are not convinced
Hogg (1977) further
it is necessary
to keep to the posted speed
involvement and speed and, in particular, the danger of large
deviations in individual vehicle speed from the average speed of
traffic.
These relationships
with an involvement
drivers and with minimum
above, the average
take the form of a U-shaped
curve
rate higher for very low and very high speed
involvement
speed (see Figure
The exact relationship
rates at, or marginallY
1.1).
of the road conditions.
Lay
was difficult
initial research because estimates
Research
Triangle
Institute
by using actual measured
In a submission to the House of Representatives Standing
Committee on Road Safety (1984), Swann claimed that people drive
to their perception
by Solomon
between crash
and,
limit.
This evidence led McLea~ (1977) to suggest the need for
alternative forms of speed enforcement.
according
for rural highways
that there are relationships
of any relationship
the speed limit and crash involvement
hence, do not believe
The basic research undertaken
(1964) illustrated
to establish
from this
of speed had to be taken.
The
(1970) extended this early research
speeds and showed the shape of the
relationship
was not neaYly as pronounced
by Solomon.
West and Dunn
as that first suggested
further and showed different
involvement
rates when inclUding
and
(1984), too, argued that motorists' perceptions have a primary
influencm on travel speeds.
If this is the case, it may be
possible to change road conditions to influence the perception of
what is a safe operating speed.
Conceivably, this would result
involvement
selecting
a lower travel
-
2 -
speed.
.
••
excluding turning vehicles.
If there are few turning vehicles,
the curve is extremely flat and there is little difference in
in drivers
~
(1971) took the research one step
rate for speeds up to 25 km/h from the average speed.
This relationship
Table 1.1.
reported by West and Dunn is illustrated
- 3 -
in
I,
ME-
~----------------_.~--------------------_._
_--_._.-
..
TABLE
RELATIONSHIP
BETWEEN
FIGURE 1.1
Involvement
travel
speed,
rate
day
!l
~.~
-.\..--'- /
Speed
Dunn,
CRASH
INVOLVEMENT
1971)
Involvement
Rate
per
Including
Turning
Million
(mph)
Excluding
Crashes
- +-
- 5.5 than
+More
to
10,000
vehicle
miles
Deviation
-~
~
AND
and
night
1964)
-"" W'LLlON
-- IN "'CC'O[IHS
100
-WIL[S
L'
I INVOLV[O
./ V[HICL[
tl ...., .... P["V[HICL[S
!lIHIIOLII[W[NT
•••••T[ 's NU•• II[
•• Of _
Of
1\. T•••••V[L
\\.
J
;\.. \\ ••••••.•..
0
_NIGHTTlME
50,000
1.1
SPEED
West
_-._~
by
and
(Solomon,
'"
(from
..
5.5
15.5
to
15.5
5.5than
to
Less
+
Turning
Crashes
42.3
6.3
2.3
.7
5,000
r.l
1.6
.8
1.5
1.0
8.5
6.9
E-t
~
E-t
Z
f>l
~
..:i
g
Z
....
1000
Munden
500
crash
(1967)
rate
Solomon.
and
Munden's
however,
as he
relatively
100
50
20
30
40
TRAVEL
SPEED.
50
60
MP
70
by these
the
those
crash
who
rates
(1971)
supported
the
by
the
crash
concluded
that
the
lower
upper
involvement
limit
and
control
rate
with
can
in
Croft
and
relationship
correlation
limit
that
reducing
people
who
rate
curves
and
or three
speed.
and
------_._~.-
_
I" -
effective
h
crash
between
from
the
between
(1971)
concluded
the
mean
rate
speed.
crash
reviewed
that
of
much
of the
there
is an
crash
and
the
In addition,
severity
and
speed
__
.....__ .__ .
literature
on
intimate
speed
they
deviation
claimed
where
- 5 ~Ioo.......-
and
speed
more
- 4 -
in
Hauer
of overtaking
upper
times
overtakings
derived
of overtaking
rate
lower
drove
complimentary.
involvement
to the
by
research,
slowly.
of traffic
both
be two
previous
still
the
between
reported
rates.
Cumming
vehicles
involvement
the
of
are
function
on highways
relation
that
relatively
crash
showing
a V-shaped
to
from
drove
researchers
related
speed
differed
studies
also
the
a V-shaped
similar
different
was
than
finding
with
traffic
limits,
speed,
the
Hauer
80
H
reported
compared
fast
Nevertheless,
10
also
relative
._~---_._------------------
a
severity
of
T
increased
with speed.
If the objective
of speed control is to
fact, argued that variance,
road deaths in the USA.
speed.
also shown that the probability of a fatality increased markedly
for speeds above 70 mph (112 km/h).
Bohlin (1967) reported that
the relationship between injury risk and speed was curvilinear
Lave (1985), in
not speed, was the critical factor in
with an increase
in probability
of the injury being greater at
higher impact speed.
Bohlin further showed that the relationship
can be changed by the wearing of seat belts.
1.3.1 Speed and the Number of Crashes
A number of countries have sUbsequently amended speed limits
on rural highways and several have investigated the effect of
reduced speed on crash involvement rate (Organization for
1.4 POSTED SPEED LIMITS
Speed limits traditionally
Economic
Co-operation
and Development,
1972; Johnston, White and
consist of general rules that
apply in discrete areas, such as urban or rural settings.
Where
these limits vary due to roadway, traffic or adjacent land use
Cummings, 1973; Nilsson, 1977, 1981; Johnson 1980; Salusjarvi,
1981; Hearne, 1981; Christiansen, 1981; Lassare and Tan 1981).
conditions,
the practice
Limits themselves
is known as speed zoninq
being the
maximum
observation
facie, being the speed above which a driver would have to prove
that the speed was not incompatible with prevailing conditions.
analysis
before and after speed changes,
of the beforevalidity
and after-crash
of the effects
lack of rigorous
data, and questionable
attributable
in the total number of crashes.
appears to have been a reduction
speed reductions as well.
Furthermore,
Cleveland
there
(1970) and Cumming and Croft (1971) reported reductions
in the number of excessively speeding vehicles when appropriate
absolute limits were chosen, although the average speed was not
in the level of injury with
affected very much.
This has not been proven yet for prima facie
limits (Herbert and Croft 1979).
Traditionally,
be expressed
of energy resulting
by the relationship
between
from any collision
can
85th percentile
vehicle mass and speed.
research undertaken
itself.
This means that increased speed increases the severity
of a collision and most probably increases the associated level
for vehicle
occupants,
although
energy between vehicle and occupants
severity of injuries sustained.
speed limits have been set at or near the
"the
85th
the exchange of
straight
on~y
drivers
-6-
by Witheford
by
and
distribution
set
choose to travel.
percenti~e
approximated
can also influence the
be
speed
steep
s~ope
the
~ower
is
of
85th
will
violators".
-7L
This choice stems from
(1970) who stated that:
that
a sf?eed ~imit.
below
a little
to
I·
speed of traffic, that is, the speed at or below
which 85% of motorists
Wadsworth (1966) showed that kinetic energy is generated by the
moving vehicle by the square of the speed rather than speed
of injuries
to travel, or prima
Absolute maximum speed limits can exert a greater influence
on the distribution of speeds than a prima facie limit.
1.3.2 Speed and Crash Severity
The dissipation
is permitted
to speed
reduction.
However, the weight of evidence suggests that, where
speeds have been reduced by speed limits, there has been a
reduction
speed at which a vehicle
,-
,
(NAASRA, 1980).
can either be specified as absolute,
Individually, these studies may be of limited value because
of various shortcomings, such as relatively short periods of
statistical
--------
Solomon (1964) concluded that this increase was extremely
rapid at travel speeds in excess of 60 mph (96 km/h).
It was
reduce the number of crashes, the aim should be to reduce speed
variation.
However, if it is to reduce crash severity, the aim
then should be to reduce the absolute
. -~~....--...
most
8ecause
the
typica~
percentile.
cause
desirab~y
of the general
speed
a speed
a large
number
limit
of
l.
1
:;;y,'\]"
construction
JOscelyn, Jones and Elston
(1970) undertook a national
survey of practices used to establish maximum speed limits,
together with a major review of the various techniques for
establishing
speed limits.
From a screening
analysis,
it was
Suggested that ibree methods were Worthy of further consideration
for full scale implementation.
These included the theory of
speed distribution skewness (TaYlor, 1965), cost orientation
(Oppenlander, 1966), and the 85th percentile method.
and width and roadside development.
index is subsequently
determined
The speed zon~
after assessing
the existence
and likely effects of these factors at a particular
The speed zone index, therefore,
acknowledges
road site.
the role of
perceptual factors in speed determination on the road. This view
is also shared by the National Association of Australian State
Road Authorities
(1980) and Lay (1984).
the method adopted
for arriving
Unfortunately,
however,
at the 65th percentile
speed
using the spe~d zone index relies upon ·several years of
Their preferred
relationship
technique
between deviation
analysis
showed a strong
of the speed of the crash vehicle
fro_ the -ean .peed of the traffic .tream (Jo.ce'yn et a' 1970).
The analysis also showed that the cumulative crash rate Was
re'ative'y independent of .peed unti, the 85th percenti'e va'ue,
after which it rose exponentially.
The study recommended that
the
observed
maximum
speed travel
limits Speeds.
should be based on the 85th percentile of
experience"
in interpreting
more objective
minimise
the possibility
1.5 DEFINING
The word
different
the likely perceptual
A more recent alternative
is a Cost-benefit
to the 85th percentile
analysis.
Cowley
speed
(1981) Suggested
'i_it.
that 'peed
transportation
cou'd be .et at 'peed. which minimi.e overa"
cost.
The European Conference of Ministers of
Tran.port (1977) discus ••• co.t-benefit ana'ysis with re'pect to
general limits, revieWing preVious studies and discuSsing the
benefit. and di.benefit. of the method.
It was concluded that
the Cost-benefit approach would have to be developed further
before it could be effectively evaluated.
1.4.1
1be
Speed Zone Ind~
A
to
of bias.
SPEED PERCEPTION
"perception"
has different
people and professions.
of an individual
interacting
connotations
for
On one hand, perception
been used to refer to the relatively
approach
effects.
and rational approach would be desirable
automatic
has
sensory processes
with his or her environment.
this sense, it is the first stage of the pSYChological
In
process
that occurs between human stimulation
and response, and can be
referred to as the sensory perceptual
staqe
Alternatively,
deliberate
perception
and conscious
(See Figure
1.2).
has also been used to describe
thought processes
an individual's
the
involved in human
response,
involving
desires.
processes
Perception here involves higher order decision making
where the social consequences of an action can
influence
the ultimate
response.
referred to as the coqnitive
beliefs, motivations
For convenience
perceptual
and
sake, this is
staqe.
There are often occasion. where free .peed studie. cannot be
undertaken to assess the 85th percentile ValUes.
In these
The speed at which a driver chooses to travel can clearly
.ituation., Traffic Commi ••ion Victoria (1976) Out'ined a .ethod
of estimating the likely 85th perCentile Speed, based on an
evaluation
of various factors that suPPOsedly
influence
the
SUbjective assessment of speed.
These factors include such
things as roadway characteristics, alignment, shoulder
involve both of these perceptual constructs.
While sensory
perception will determine from the outset what information is
available to a human operator in a partiCUlar stimulus situation,
the internal states or social forces can nevertheless influence
the form of the ultimate
response
to that information.
- 8 ••. "~7""'" " .,
·'r.·':':'",'!"I •.
':~fi"'l.:~'r
•.
- 9 ".':"1
•.••
_ ••
~
:'IJJ:,'
w.,
w·,..v
I
Given that sensory perception
resPonse
in his or her environment,
is the basis for a human
manipulating
the visual CUes
1.6.2 AbsQlute Judqements
Of S~eed
involved in sensory perC~tion
on the road could lead to longterm improvements in road behaviour.
This project is
specifically
aimed at testing
sensory perceptions,
There have been a few studies which investigated
and
ability to estimate his or her own travel speed.
concentrates
on the effect the driving
input processes
1.6 PREVIOUS
of a driver's
RESEARCH
jUdgement
environment
has on the
of sPeed on the road.
found that speed judgements were under-estimated
responding to films and by passengers
situations
IN SENSORY PERCEPTION
There is only a limited number of studies that have
investigated the sensory aspects of Speed perception.
These are
summarised below, along with a theoretical account
of the role of
movement in the sensory perception of speed.
1.6.1 Visual Performance
under-estimated
in actual traffic
Salvatore
(1968,
(1974) reported that subjects consistently
speed across a 20 to 60 mph (32 to 96 km/h)
velocity range in both vehicles and simulators.
and Salvatore,
(1963)
by subjects
in the medium and high speed ranges.
1969) and Reason
a driver's
Hakkinen
furthermore,
Both Hakkinen
showed that errors of estimation
increased sUbstantially when sensory inputs (visual, auditory,
kinesthetic, tactile and vestibular) were withheld.
Studies
Evans (1970a, 1970b) found that slow speeds were, in fact,
well under-estimated
Denton
(1971, 1973) reported
that speed jUdgements
in a
driving simulator and on the road Were highly dependent on the
nature of the visual pattern presented to the driver's eye.
Transverse line treatments Were introduced at selected
under-estimated
that the perspective
in the United Kingdom
to induce drivers
to slow down
during their approach to these intersections.
Evaluation studies
(Denton, 1973, 1976; Rutley 1975; Helliar-Symons,
1981) reported
Subsequent reductions in Speed, SPeed variation and lateral
view presented
tests of speed estimation
judgements.
roundabouts
whereas high speeds were only slightly
in vehicle tests on the road.
was critical for replicating
Any test of speed perception,
take account of the absolute
presentation
Moreover,
he found
to sUbjects in laboratory
therefore,
road speed
needs to
level of speed involved and the
method of moving road environments.
1.6.3 Retinal Streaminq
POsition,
disSipate
although the speed effects of this treatment tended to
somewhat with time (Denton, 1973; Rutley, 1975).
The concept of "retinal streaming" was proposed by Gibson
(1950, 1958, 1968) and Calvert
More recently, Cairney and Croft (1985) and Cairney (1986)
reported two studies which investigated the effects of various
environment and road factors on drivers' speed jUdgements
cues used in perceiving
(1954) as an explanation
speed.
of the
In essence, retinal streaming
explains how the visual pattern presented to a moving observer
varies from a stationary
image at the point of fixation of the
Unfortunately, these studies used static Photographs of roads and
SUbjects made verbal speed limit responses; the reSUlts,
eye (the fovea on the retina) to a blur of increasing magnitude
therefore, are diffiCUlt to interpret oolely in term. of Oenoory
PerCeptions of SPeed.
Nevertheless, with slight mOdification,
provides a means of interpreting
the further the distance
from visual stimulation
from the fixation point.
velocity
in the periphery of the eye.
theinvestigating
approach adopted
these
studies could prove a Useful means
Of
road in
Speed
perception.
- 10 -
r---
._._
_.-,¥._._.",~
""'~"~7'~~:,-..,. .•.
~
--., .• ---
Thus, it
information directly
11 -
i
•.- ~i ~k_~
~ ;•.•.
i"...
:J..
1
_._-----------------+ ••_-----------_ .._-.._-~._
-""'
..,._."-~_
..- :--The notion
of "retinal
been criticised
Beverley,
form of relative
Crook,
However,
2.
has
1938; Gordon,
cue for the perception
1966; Moore,
and Wilkins,
Gordon
There are many possible
Jennings and Demetsky
of speed (Gibson and
1968; Lee and Lishman,
variables
1977;
(1968) described
the consequences
image on the road in terms of velocity
retina of the eye.
They argued
gradients
that the perception
of
variables
of speed perception
(1950) motion
parallax
are in general
concept
on a driver's
perception
that are
of speed.
that they claim will influence driving behaviour
A detailed
undertaken
literature
on the
review was
to indicate the likely effects of these
for speed perception.
on the
TABLE 2.1
of velocity
POSSIBLE
on the road can be determined solely from the integration of
movement information on the retinal surface.
These theoretical
accounts
factors in the environment
(1983) listed 16 road and environment
road (see Table 2.1).
1980).
(1966) and Moore
CUES FOR SPEED PERCEPTION
likely to exert some influence
subsequently
a moving
POSSIBLE
1973; Regan and
most authors agree that some
coding on the retinal surface of the eye is an
important
Harrington
in depth perception
(Johnston, White and Cummings,
1978; 1982).
extremely
streaming"
ROAD AND ENVIRONMENT
& Demetsky,
(from Jennings
VARIABLES
1983)
accord with Gibson's
in depth perception.
ROAD or ENVIRONMENT
FACTOR
time of day
roadway grade
radius of curvature
length of curve
lane and road width
existing pavement markings
shoulder width (if any)
nature of adjacent
intersecting
- in curve
lane
roadways or driveways
- before curve
- after curve
I"
traffic volumes
b,
average speeds of traffic
delineator
spacing
(if any)
delineator type and number
condition of delineators
weather
conditions
sight distance
-
-------
-.-
..
12 -
- 13 --
•......
(if any)
.
--_
_-~--------_..•__ ....•_ ............•.•••••••••••......••••••••............
-------------- ---"--"-
..
The length of the trip and the number of passengers
2.1 DATA SOURCES
vary between rural and urban journeys.
Hirsch
The evidence
influence
of road and environment
speed speed perception
data sources,
factors likely to
was drawn from three separate
namely crash data, free speed data measured
road, and performance
data from road and laboratory
The fi~st two categories
involved
mainly
on the
testing.
correlational
establish
definitive
The third
category
experimental
causal relationships
of data includes
from these studies.
both correlational
and
results and was more likely to be relevant to this
study.
(1966) reported that trip distance and purpose
different
studies.
samples of drivers and environments
likely to induce speed adaptation
In any event, the findings
with rural travel are more
and driver fatigue.
Mast,
Jones and Heimstra (1966) showed that driver tracking error and
speed control were significantly different in the last hour of a
six hour continual tracking task.
Safford and Rockwell (1966)
reported that speed control diminished with time over a 24 hour
from all three data sources were
These data probably
Triggs
prolonged
2.2 URBAN AND RURAL ENVIRONMENTS
between
these two environments.
different
(1986) described
speed adaptation
exposure to vehicle
practically
generally
reflect pure driver fatigue influences.
First,
is likely to
speed limits are
for urban and rural roads.
Road design
incorporates proposed speed limits and other major urban-rural
environment differences.
Hence, urban and rural roads differ
considerably
in terms of their engineering
expectations
(Lay, 1964).
and community
and Rockwell
(1961) and Jennings
and Demetsky
(1963) noted that travel on rural roads was noticeably
different
from that on urban streets in that speeds were generally higher,
traffic volumes were much lower and the severity of crashes
increased.
Sanderson
and Corrigan
(1984, 1966) also demonstrated
that many drivers disproportionately
than rural limits.
-
violated
urban speed limits
14 -
stopping when slowing down to pass through a country
not necessarily occur with driver tiredness.
Nevertheless, it
does tend to occur only after long periods of fairly constant
high speed travel (Reason, 1974) and thus is often compounded
with driver fatigue effects.
the role of the road and its
surrounds ih speed perception,
therefore,
should be concerned
with the influence of urban and rural environments.
"urbanness" and "ruralness"
Moreover,
are not discrete categories,
rather end points on a road environment
continuum,
but
it would be
useful to consider a range of road environments for speed
perception comprising various combinations of urban and rural
features.
-
-.------"'-
It is the feeling of
village, even though travel speed is still substantial.
Speed
adaptation differs from driver fatigue in that adaptation does
Any study which addresses
Hungerford
as the effect where
speed causes SUbjects to under-
estimate their speed jUdgements.
There are several reasons why speed perception
....
used in these
driving task, but interestingly, a rest period of only a few
minutes was sufficient to offset these effects in the short-term.
only used for evaluating which variables were likely to be of
interest and for generating hypotheses for further testing.
differ
can also
Joscelyn et al (1970) and
significantly affected speed, although the number of passengers
gave conflicting results.
This may have resulted from the
The longer journeys associated
studies, where environmental effects were deduced from vehicle
behaviour observed on the road.
Hence, it is impossible to
"-""
15 -
as
r
~-_.
----,
~-
.0·
2,3 ROAD ALIGNMENT
In a review of the relationship
between road geometry and
single vehicle crashes, Sanderson and Fildes (1984) reported that
the most consistent associated road feature was horizontal
curvature (see Table 2.2).
In most instances, decreasing the
...
...
f"X in 0run-off-the-road
>
radius of curvature led rn.ous
to an increase
0
~
c
~
crashes.
I1cBE:.AN
W"RICHT
RAFT
unexpected
,
:tADOR
Studies which evaluated
Some studies
although
superelevation.
1198
Sanderson
vertical
suggested
The results
dependent
in Victoria
further
and Fildes
between
(Raff 1953).
curvature effects were
a relationship
for gradient,
with road
however, were mixed,
technique
employed.
More
on.
(1984) analyzed
single vehicle crashes
1978 and 1982 and found that 47 per cent
on bends and 53 per cent on straights.
crashes
have also been reported
They concluded
on curves were over-represented
in the statistics, crashes on straight
significant problem in this Australian
roads were still a
State.
Similar results
for other Australian
1970; Peter Casey and Associates,
States (Cowl and
1979).
YES
The performance
differences
sections
~
~
:2
was found for camber or
on the measurement
that while single vehicle
Fairlie,
~
1,
no relationship
will be said of gradient
occurred
-
...
there was some suggestion that an
sharp curve may be crash inducing
inconclusive.
seemingly
""
j
c;;"--z0"~
"~
c-- - (UBI
z"--~~
z"
" ...::!
". 0
.•a""
~
3::!
z u
~::!
M
'"
YES
SOM.E
HO
...
SOM.!:
HO
YES
YES
...
Z
f-HO
...
~>>
...1980)
0
~<
>u
"'
fu...
~<
xu
Of0
uYES
fZ
SOM.!:
HO
YES
YES
SOM.!:
YES
SOM.E
YES
,f19821
HO
SOM.E
YES (19S)I
(1'81)
7S1
((19
;;;<5~
The frequency
of{19'OI
curves did not appear to influence
C;UPTA
JArH
COOPER
KALL
COWL'" ,, IADOR
t"AIRLl£
ACOlT
DEDi
crash rates, although
contour,
""""""'"
bu,
in driving
of roadway.
and Shinar, McDowel1
differences
literature
between
and Rockwell
relat'ionship
I,
!
(1975)
(1977) all reported significant
in driver eye movements
between
Table 2 _2 Summary of accident cilusation studies \·,hich
implicate road geometry with crashes (from Sanderson &
Fildes,1984).
straight roads and
curves.
They noted that visual fixations for straights was much
more static and generally involved longer distance fixations.
••~." ~·~"""'_~'4'·
~"'~1:'_1;,?"~",,
no
straight and curved
(1966b), Blaauw and Riemersma
-
NO ""
confirmed
also suggests considerable
performance
Gordon
r-elationsh'ip
16 -
-
L
_.
__
17 -
T
.-'-".
- ...
~-,"-
!
Triggs, Harris and Fildes (1979). Fildes (1979) and Triggs,
Meehan and Harris (1980) reported differences in delineation
had a strong crash effect.
requirements
likely relationship
involvement.
for roads of differing
perceptual
vertical and horizontal
curvatures.
They argued that these differences reflect differing
visual requirements between crests and sags in the road. Fildes
(1987), in fact, demonstrated
geometry
as a vehicle
differences
approached
in the perspective
effects
relationship
the visual demands in the approach zone of a road curve (straight
section) are noticeably different to those in the curve itself.
on driving
Straight
needs to take account
sections
of differing
appear to be visually
and would be a logical starting
data here is much more relevant.
in several studies
(Oppenlander,
Appleyard,
Moreover,
1981).
road alignments.
less complex for drivers
direct relationship
point for research in this area.
which encompasses
The crash data on the role of the number of lanes and lane
Mc Lean and Hoffman
different
crash rates on two-lane
(1970) suggested
curves in the USA, although he failed to
Peter Casey and Associates
evidence
lanes on straight
suggestion
concerning
sections
of an increase
(1979), however,
relative
Evans
found no
A correlation
Solomon
and four-lanes
head-on
no consistent
collisions,
relationship
although
...
.•
~.'.::.;':~~
their
However,
this is probably
than street length per se.
(Oppenlander,
McLean
1966; McLean,
1980;
related
The smoothness
related to vehicle
1982) but not crashes
(Raff,
(1982) also reported mean free speeds about
roads than on equivalent
15-20
paved
evidence then of an association
between road category and lane width with free speed and crash
involvement.
Given that variations in lane width and the number
or absence of a median
- 18 '~~\O'.~
(1981).
There is considerable
between median width and
the presence
et al
speed and
(Leong, 1968; Loder and Bailey,
km(h lower on unpaved two-lane
surfaces.
less
than 15 feet (5.4m) wide were twice as likely to have a head-on
crash than four-lanes greater than 15 feet (5.4m) wide.
He
reported
Joscelyn
roads adjusted
speeds in the slower parallel
of the road surface was found to be directly
tended to have
roads had a head-on crash
sections,
for narrow lanes at high speeds
lane widths and speeds.
was also suggested between vehicle
more to sight distance
speed
(1964) found that two-lane
surrounding
(a perceptual
feature).
with vehicle
Smith and Appleyard,
1953).
rate five times that of four-lane
speed and "apparent width"
of the immediate
that drivers in multi-lane
street or block length
However, there was some
and motorways
road
(1981) report a
lane, although they only reported data from other studies to
support this claim.
in fatal crashes on straight narrow
roads of sub-standard rural widths,
less fatals than other highways.
for 11'3"
(1972) found that drivers adopted
strategies
speed in accordance
safety and number of
in Australia.
steering
than for more moderate
(1985) argued that fatality rates in the USA differ by a factor
of four or more depending on the type of road involved.
conclusive
between drivers'
the influence
of a geometric
widths reveals some interesting trends.
Raff (1953) reported
that wide pavements and shoulders were associated with lower
More recently,
1966; Smith and
(1968) found greater
smith and Appleyard
on the actual road surface
interpretation
on tangents.
1966; Leong,
Vey and Ferrari
and less speed and headway variance
bridges in Philadelphia.
AND LANE WIDTH
find any strong association
A positive
(3.4~) lanes over 9'9" (3.0m) lanes across two comparable
influence
environment
2,4 ROAD CATEGORY
and crash
was reported between vehicle speed and street width
speed, less headway
Hence, any study of the likely road or environment
to ascribe
because of the
between lane encroachment
The performance
road
a road curve and claimed that
It is very difficult
from these data, however,
~
19 -
I
L
t,
1
•...
. -~,---- t
~
.
'.
,
...•
-
---
-,..
.-_ ...
I
i
and type of lanes has a marked
perspective
view presented
is quite probable
influence
would appear that the major benefit for edge lining on straight
on the geometric
to a driver moving along the road, it
that these features
will also influence
speed
sections of two-way
1986; Cottrell,
perception,
rural roads is for maintaining
position within the lane itself
1985).
Evidently,
edgelines
While
define
AND DELINEATION
road markings
lane width,
improving
and delineation
are normally used to
of the road ahead.
(1979) and Godthelp,
Milgram
Gordon
and Blaauw
Fildes,
(1966),
(1984) argued
1979; Triggs, Meehan and Harris,
and Smith,
1985), night-time
that the edgelines
of the road are used by drivers to control
1983) and curve negotiation
vehicle
1980).
of line marking
speed and travel path.
may well influence
perception
Thus, the type
of the road ahead.
influence
straight
witt and Hoyos
(1976) reported
in the approach
adopting
a more suitable
in a vehicle
reported
simulator.
speed profile
Rockwell,
that novel pavement
with time
may be site specific
(Hungerford
in drivers
while negotiating
Malecki
markings
speed and roadway width on curves,
influences
that a varying pitch broken
to a road curve resulted
and Shinar
a curve
(1974) also
can influence perceived
although
they noted these
and might not necessarily
and Rockwell,
found which specifically
Symons
Denton
transverse
(Hungerford and Rockwell,
However,
evaluated
(Godthelp and
there were no studies
the effects of guideposts
time curve negotiation performance.
Nevertheless, the relative
movement characteristics of guideposts in a driver's view of a
roadway,
influence
suggest it could be worthwhile
of guideposts
on the perception
speed could be modified
AND GRADIENT
Road design assumes that sight distance
by
in setting speed limits
stripes across the road in the approach zone to
markings, then, seem to influence speed judgements
road sections and at intersections.
were
for curved
Leong
and sight distance
Schneider
distance
Lum (1984), however, found no effect of narrow longitudinal
pavement markings (with raised pavement markers) on either the
on straight
residential
sites.
on straight
roads for wide edgelines
sections of road at
Others also have shown no difference
in mountainous
areas
(Cottrell,
in speed
1985) or special
(Garber and Saito, 1985).
It
is a critical
(Joscelyn et al 1970, NAASRA
(1968) reported a positive
correlation
while Olson, Cleveland,
between
and crashes.
However,
therefore,
1980),
Fancher, Kostyniuk
and
sight
this latter study did not control
(Galin 1981) and,
the results must be treated cautiously.
In his review of the spot speed literature,
(1966) placed sight distance
important
factor
spot speed
(1984) also noted a similar relation between
for other speed factors like traffic density
treatments
testing the
of speed.
(19BO) and Helliar-
intersections.
However, only the speed variation reductions
still evident at sites 18 months after installation.
Road
mean speed or speed distributions
on
19BO).
(1974), Agent
(1981) found that perceived
1985).
(Neissner,
posts and post height can
road guidance at night
1982; Blaauw,
2.6 SIGHT DISTANCE
As noted earlier,
strategy
crashes
vehicle performance during the day, presumably because these
delineation treatments are mainly reserved for improving night-
moving
hold
1980; Nemeth, Rockwell
run-off-the-road
Under some circumstances,
Riemersma,
edgeline
view of the road
to the driver.
heading
direction,
sections
Guideposts with reflectors have been shown to have a marked
influence on curve detection (Triggs, Harris & Fildes, 1979;
they are also adopted more generally for
the visibility
Riemersma
presented
1979; Triggs,
on straight
do not really add very much to the perspective
2.5 ROAD MARKINGS
a safe
(Triggs and Wisdom,
variables,
compared
claimed that sight distance
- 20 -
to traffic density.
restrictions
- 21 ..,._.--, ..
Qppenlander
amongst the significant
~-,.,-------~-_._",
but less
McLean
induced a small
(1982)
..
reduction
in the speed adopted
by the faster travelling
drivers,
but had little, if any, effect on the speeds of other drivers.
However,
he argued that it is difficult
distance
effects
from other effects
in the study.
to separate sight
Traffic volume, however, does seem to exert a marked effect
of the changing road
on the speed behaviour of drivers on the road and is likely to
have some influence on a driver's perception of the road ahead.
geometry.
In an early crash data studY in the USA, Raff (1953) found
no correlation
between
grade and crashes.
looked at a range of geometric
accurately
controlled
variables
for gradient
This study, however,
speed by Oppenlander
effects alone in road crashes.
(1966), but Troutbeck
differentiate
it is difficult
sight distance
it from sight distance
to separate
observed
spot
not able to
(1976) was
effects.
The National
In short,
the effects of qradient alone from
Association
(NAASRA, 1980) suggested
influence
and could not have
Road gradient was cited as a factor in influencinq
clearly
This may have been due to minimum traffic volume levels observed
speed.
there was no relationship
on free speed
of commercial
State Road Authorities
the nature of the traffic mix can
However,
heavy vehicle volumes
percentage
of Australian
vehicles
observed
for
(Duncan, 1974), or for the
on crash rates (Raff, 1953).
Generally, there appears to be a lack of conclusive evidence on
the effects of the traffic mix and further research is needed for
clarification.
in the speed literature.
2.8 NIGHT AND DAY
All things considered,
there does appear to be some evidence
that the amount of sight distance
speed judgements,
relative
although
and gradient may be a factor in
there is some dispute over its
importance.
2.7 OTHER TRAFFIC
Night and day driving conditions represent extremes
driver's visual information continuum.
The illumination
at night are some 200 times less under headlamps than daylight
conditions and background information is almost completely absent
(Fildes, 1979).
AND DENSITY
higher at night
speed.
varying
Traffic
vehicle
1974; Armour,
traffic
volume
and density,
and in the opposite
spot-speeds
1983).
be a compounding
have been associated
(Oppenlander,
In general,
leads to reduced
both in the direction
direction,
travel
1963, Rankin
increasing
(OECD, 1980; Joscelyn
of
with
drivers
(Drummond,
flow density
can
volume on straight
1979).
However,
roads
with increasing
(Raff, 1953; Peter Casey
traffic
& Associates,
these studies did report a threshold
et al 1970) and those involving young
1985).
effect at
OECD (1980)
reported
large speed differences
at night in bad
abnormally
accurate
Triggs and Berenyi
judgements
crashes on freeways in Melbourne
ceiling.
road delineators.
supporting
- 22 -
I
\
speed at night.
They attributed this to the increased angular speed of elements
visible to the driver which, under headlights, are much closer
increase,
any threshold
I·
(1982) found that subjects made more
of rural road travelling
than normal and form streaming
he did not observe
with different
during night conditions.
high volumes, presumably because traffic flows became severely
restricted.
Worsey (1985) also found a similar crash rate
although
and
driving performance
weather.
increased
speeds tend to be
Co-operation
A number of factors have been associated
factor here.
Crash rates generally
though, driving
for Economic
& Hill,
the amount of
speed, although
Interestingly
(Organization
Development, 1980). This may help to explain the high propensity
of fatal and serious injury crashes that occur during darkness
The Australian Road Design Handbook (NAASRA 1980) states
that the volume of traffic will influence a driver's chosen
travel
in a
levels
Sanderson
patterns
(1985) reported no difference
in
that were both lit and unlit,
the notion of perceptual
- 23
~_.
produced by reflectorized
narrowing
-
at night.
T
Elliot
(1981) surveyed
to formulate
hypotheses
found that drivers
various
regarding
were actually
groups of Australian
drivers
their speeding behaviour.
encoura~ed
He
".-.-....
-.""~""differing
results on the significance
and condition
of road shoulder, width,
in speed determination.
to travel at higher
speeds at night because the road somehow seemed safer.
verbal reports are inherently unreliable for perception
However,
studies
Early resear.ch such as that cited in Joscelyn
and Leong
correlated
et al (1970)
(1968) found shoulder width and condition
to free speed.
In reanalysing
to be
Leong's data, Troutbeck
as many of these influences occur without the driver's knowledge.
The role of night driving in speed perception, then, requires
further investigation, although initially, perhaps not as
and claimed Leong's result was spurious.
important
crash rates and shoulder paving was found by Turner, Fambro and
as the effects
2.9 ROADSIDE
of the road geometric
variables.
(1976) however, was unable to find any effect for these factors
A correlation
Rogness (1981) when they applied 4 more rigorous categorization
of variables than that used in the earlier studies.
The
DEVELOPMENT
rationale used by Turner et al was that road shoulders
Roadside
development
can be broadly defined as any aspect of
the environment close enough to the roadway to influence driving.
NAASRA (1980) claimed that the roadside environment will
influence
traffic
speed and this has particular
traffic engineering
and planning
(Brindle,
importance
1980).
for
and perceptual
effects
and pedestrian environment.
Jaenichen, Ernest (1984).
occur in
such a variety of types that careful classification is necessary.
However, care must be taken when using shoulder crash studies to
assess speed as different
so on may have different
types of shoulders,
crash potential
paved, gravel and
in themselves.
The "Woonerf"
(building total environment) in the Netherlands, for instance,
uses benches, different pavings, winding vehicle paths and other
physical
between
to reduce vehicle speed in a car
(Refer also Boeminghaus
Bever-
Rural roads with no prominent
side features have been
associated with lower speed judgements than similar treelined
roads (Shinar, McDowell
&
Rockwell,
1974).
Triggs
(1986)
suggested that this effect may be due to either increased
peripheral
stimulation
or motion parallax effects.
Whatever
the
case, it is apparent that this variable needs to be considered
Wilson
(cited in Joscelyn
culture and establishments
et al 1970) found that roadside
resulted
in lower spot-speeds
on four-
fully in any investigation
of speed perception.
back would also seem to warrant
Building
further study, although
set-
the
lane highways.
Smith and Appleyard (1986) also reported that
house setback distance was positively correlated with urban car
evidence to date does not suggest the same for shoulder width.
speeds.
2.10 DRIVING EXPERIENCE
operating
important
They hypothesized
that an urban
·walled in" effect was
on the driver and considered building type to be an
consideration in urban speed behaviour.
I
Evans and Wasielewski
(1983) reported a direct relationship
between the amount of acceptable
headway and youth, while Cowley
Rankin and Hill (1974) did not find land use alongside their
test roads to affect travelling speed, although it was noted
(1983) linked speeding with young males.
Free speed surveys by
Seal and Ellis (1979) and Wasielewski (1984) in the USA show that
worthy of more detailed
younger drivers tend to travel at faster speeds.
investigation.
land use to be a significant
variable
due to the very broad based categories
control for factors
variable
Part of the failure of
in this study may have been
used and the failure to
such as traffic volume.
This lack of
control was, in fact, noted in explaining
their
was not observed
(Johns, 1981) or Victoria
drivers are constrained
this
(Manders, 1983) where inexperienced
to lower maximum speed limits.
these results are not directly
Thus,
comparable with the overseas
findings.
- 24 -
However,
for first year drivers in South Australia
- 25 -
\
'--
Driving
perception
experience
has been found to influence the
of rOad hazards
drivers also responded
(McKeown,
differently
1985).
These differences
Inexperienced
the road conditions
(slower) to dangerous
nature, it is difficult
situations
strategies
(Quimby and Watts, 1981) and adopted less safe visual
(Mourant and Rockwell, 1972).
In controlling
deviations
in straight
line driving,
Riemersma
(1982) explained
observed differences as due to the use of different strategies
cues.
young and experienced drivers in making use of lateral Speed
braking and cornering
by
experience may have perceptual consequences for Speed assessment
on the road.
It would be worthwhile including this variable in
any SUbsequent Speed perception experimentation.
quantitatively
on the
on slippery roads.
Inclement weather may
also interact with other road conditions
(such as night vision)
leading to an increase
in road crashes for night driving
in wet
(OECD, 1980).
Overall then, there is some suggestion
that weather
conditions may unduly influence a driver's
although weather needs to be carefully
experimental
2.11 WEATHER
to measure weather
of
By its
road. Hence, the results could reflect reduced visibility as
well as altered handling characteristics of the vehicle when
conditions
These results then suggest that the amount of driving
may be due, in part, to the variability
examined by these researchers.
investigations.
speed perception,
controlled
It is conceivable
in any
that the
perceptual image may be adversely influenced by the amount of
light available during inclement weather and this relationship
needs to be tested further.
Morris,
Mounce,
had a substantial
a rain simulator
Button and Walton
effect on the visual
during both daYlight
(1977) reported that rain
performance
of drivers in
2.12 PARKED VEHICLES
The degradation in performance was a function of the rain rate,
drop size and vehicle speed but not affected by wiper speed above
50 cpm. Unfortunately, it is difficult to interpret this study
behaviour.
in terms of the likely on-road effects of weather on speed
The evidence
(1977) found that SUbjects
assessed
Poor weather
conditions as less safe for driVing on a SUbjective scaling test.
This result has instinctive validity and is Supported by the
higher proportion of fatal crashes recorded during wet Weather on
highways in New South Wales Over the periOd 1968 to 1977 (Peter
Casey & ASsociates 1979). However, these effects may not be
wholly
Speed related.
Oppenlander
(1966) reviewed
studies which
found inclement weather in aSSOciation with lowered Spot-speeds,
although Olson et al (1984) found no significant differences
between Speeds at selected sites on wet and dry days.
available
on the effects of parked vehicles
pedestrians
on speed behaviour
convincing.
Loder and Bayly
parked cars and the presence
safety.
Levin
AND PEDESTRIANS
and night time Conditions.
However,
(1980), for instance,
of pedestrians
they only presented
Smith and Appleyard
correlated
strongly influenced
Joscelyn
that
a small amount of evidence
(1981) showed that vehicle
of how
I
speed was
with "apparent width" which they argued was
by the presence
of parked vehicles.
However,
et al (1970) claimed that Objects on road shoulders
had
little effect on free speed unless the lane width in total was
less than 20 feet (6.2 m).
Howarth
Surprisingly,
Thompson,
(1985) reported that driving behaviour
marginally
modified
in the presence
children had no effect on vehicle
reductions
were observed
Nottingham
in the UK.
Fraser and
was only
of pedestrians.
speed, although
Roadside
small
for large groups of pedestrians
- 26 - 27 ........
suggested
are a major threat to
showing an increase in the crash rate and no suggestion
these factors influence vehicle speed.
positively
and
and crashes is not particularly
in
I
1
Samdahl
(1986) evaluated
the effectiveness
of a
neighbourhood road safety campaign in New South Wales, Australia.
He reported instances of minor speed reductions in particular
local streets following the intensive campaign to reduce
pedestrian
casualties.
However,
they acknowledged
that these
speed changes may have been due to other factors such as weather
or parking which were outside the control of the survey. More
alarmingly
though,
they also reported
drove closer to the children
after the programme
required
to determine
SUMMARY OF THE EFFECTS REPORTED
than before,
futher confirming that the presence of pedestrians
little influence on driver behaviour.
More work is clearly
TABLE 2.3
instances where drivers
ROAD AND ENVIRONMENT
seems to have
some
some
unsure
unsure
some
some
med
low
med
low
A unsure
B
high
h
igh
high
yes
Cunsure
some
yes
some
unsure
environment
P
riority
yes
category
yes
yes
yes
(light
&
pedestrians
conditions)
yes
alignment
cars
posts)
experience
(guide
density
Factor
j,
the likely
effects of parked
vehicles
and
pedestrians on a driver's speed on
Weather
Traffic
mix
Gradient
Delineators
Parked
markings
Lane
width
Road
Day
and
night
vision
Roadside
Slight
distance
development
Driving
the road, and
especially whether the two interact in the
perception
of Speed.
Urban/rural
\
2.13 SUMMARY
The review of the literature
environment
factors on crashes
some interesting
findings.
on the effects of road and
and speed behaviour
Table 2.3 summarises
has produced
these effects
and assigns priorities to those factors likely to exhibit sensory
effects in the perception of speed.
There are TEN road ~r
environment factors of immediate prime interest:
Urban and rural enVironments, as well as combination
environments of urban and rural features.
Road alignment
Road category
inClUding
straight
such as freeways,
Lane
width,
especially
or
wide
travel
lanes.
and curved sections.
arterials,
local streets.
Sealed and unsealed
inclUded for rural environments.
Collectors
or
roads Could also be
roads that have exceptionally
A
crash statistics
B
performance
C
perceptual
studies
studies
narrow
- 28 '>.~"'~-~'-'+
r
- 29 -
--.-_.
~--',
""'f"'"",,~~'n
•... '~, "'_.~., ..•..••• ~
«
~.~U_
OF THE
FACTORS REVIEWED
3.
Roadside development comprising
in" and spacious effects.
Driving
SPEED AND PERCEPTION
those offering both "walled
As noted previously,
have attempted
experience
year of driving
experience.
THE MEASUREMENTQF
and, in particular, the effects of first
compared to those with considerable
only a handful of studies published
to measure the sensory perception
majority of speed studies in the literature
assessing vehicle behaviour
of speed.
concentrated
on the road which compounds
effects with decision making processes.
The
on
sensory
Thus, it is difficult
to
interpret these studies solely in terms of sensory perception.
Traffic density
travel speed.
below levels that severly restrict free
A review of the literature,
and problems associated
Sight distance
headway.
in relation
to road geometry and traffic
perception,
was undertaken
testing speed perception
Parked vehicles
and pedestrians
of all ages.
involving the various methods
with the measurement
of speed and
to arrive at a suitable means of
for this project.
3.1 ON-ROAD STUDIES
Day and night vision.
On-road measures generally have the advantage of high face
validity in road experimentation because they evoke normal
driving stresses.
They are, nevertheless,
perceptual
explore.
The former, in particular, could offer a relatively
inexpensive and quick solution to particular hazards for the
difficult to assess using dynamic output measurement.
perception
of speed, if shown to be an effective
countermeasure.
Except for particular installations, road markings, the type of
traffic mix and gradient effects by themselves are less
interesting variables for the perception
of speed on the road.
investigations
disadvantaged
Other factors, such as guidepost delineators, the weather
and the amount of daylight may also be useful variables to
driving skill responses.
Moreover,
In other words, input relations
in a particular
trial is much more difficult
to achieve on the road with differences
traffic.
in light, weather, and
Hence, on-road tests are always more likely
to give spurious findings in perceptual
studies.
and Forbes
(1972) discuss many of the advantages
associated
with both laboratory
Many researchers,
in a laboratory
behaviour
in the laboratory
1977).
Chapanis
(1967)
and problems
and field experimentation.
then, have opted to study road perception
environment,
(McRuer and Klein,
Noguchi,
are
control of the various factors likely to influence
driving performance
surrounding
in
in that visual effects are mixed with
and have found that certain
can predict behaviour
1975; Witt and Hoyos,
on the road
1976; Evans,
Of course, any experimental
1970b;
research needs to be
ultimately validated on the road using driver performance 'studies
of some kind.
In this respect, the laboratory experiment can be
- 30 -
-
31 -
-----..~--------~--~-~_._-.<-_.~-----
an important
first step in understanding
perception
driving
and, therefore,
for
improvedinroad
safety.
3.2 SPEED ASSESSMENT
Denton
laboratory
perceptual
experimental
(1979) argued that
simulation is a suitable means of investigating
aspects in driving.
Denton found that road
could be accUrately
belt with the subject
possible, providing care is taken to ensure the road perspective
view is accurately presented to the subjects.
In any event, it would be worthwhile conducting preliminary
trials on the road and in the laboratory to validate any
IN THE LABOROTORY
(1971) and Watts and Quimby
deceleration
conveyor
the role of sensory
has practical applications
represented
Watts
and QUimby further reported that a car body response capsule with
movie film presentation gave identical risk assessments of
The choice of whether to use static or dynamic
materials
in laboratory
dependent
on the particular
presentations
particular
realistic
simulator.
road sites as those obtained
road noise and vibration
in road vehicles when
were also included in the
While the car body would no doubt help to create a
realistic driving atmosphere, neither of these two authors showed
for
tasks.
any their
real need
for this elaborate and expensive Subject console
direction
Wierwille
(1975) and Noguchi
judgements
(Triggs, Harris and Fildes,
(Fildes and Triggs,
(Fildes and Triggs,
judgements
(1977) all investigated
made on the road, providing
the effects
care was
taken to replicate the perspective road view (Salvatore, 1969;
Evans, 1970b).
When aUditory and vibration inputs were withheld
in some of these studies, over-estimation tended to increase
(Salvatore, 1969; Evans, 1970a; NogUChi, 1977). MCLane and
Wierwille (1975), however, did not find any statistical
are based predominantly
in the laboratory
1979; Fildes
assessment
on two-dimension
amenable
static
to static
(Fildes, 1987).
(1985) and Cairney
the speed distributions
road would validate
perception.
in a driving simulator when
the stationary
However,
was necessarily
on the road.
they believe
(1986) argued that
of judgements
of static road
with free speed distributions
presentation
on the
method for speed
one would never know if the laboratory
represent~tive
of the way drivers perceive
It may be that subjects are just responding
is required of them in the laboratory
perfectly
for other pSYChological
data
speed
to what
environment.
there is no reason to expect sensory perceptions
speed to correlate
of
with free speed data on the road,
given that drivers may moderate
their speed behaviour
on the road
or social reasons.
These studies, then, suggest that laboratory presentation
may be a suitable method for investigating sensory speed
It was argued earlier that the perception
perception.
While a full simulation involving realistic road
noise and vibration would be preferable to replicate road
predominantly
processes
conditions
perfectly,
a less realistic
r-
It was argued that these
on the road, and therefore,
Cairney and Croft
matching
Besides,
differenceCues
withwere
their
aUditory
not speed
given.estimates
Static
1980), curve exit angle
1982) and curvative
1983, 1985).
scenes in the laboratory
and estimates
task employed.
(1970a, 1970b), McLane and
of laboratory simUlation on speed estimates.
In general, high
consensus was found between movie film jUdgements in the
laboratory
experimental
stimulus
is ultimately
of road curves has been used for road curve
assessments
simulation
(1968, 1969), Evans
based road experiments
1979; Triggs, Meeham and Harris,
glimpses
Salvatore
in speed perception.
3.3 STATIC AND DYNAMIC DISPLAYS
by a moving road
seated in a car simulator.
method
simulation
may also be
of speed
involves motion parallax and other dynamic
(see section
1.4).
Speed assessment
account the implicit motion cues available
32 - 33 -
retinal
needs to take into
on the road.
Indeed,
I
I
',=
in a reCent review
of the SPeed esti.ation
literature,
Trigg,
3.5 SPEAD MEASUREMENT
found that al,ost all of the studies reported involved a
moving rOad environment.
It seems inconceivable, then, that a
IN THE LABORATORY
(1986)
single static presentation
the reali,tic
assess.ent
of road scene, Would be 'Ufficient
of speed perCeption
fo,
in the laboratory.
3.4 FREE SPEED DATA ON THE RO~
There are at least three different
for assessing
criticised earlier as a means of validating their static slide
presentation method, it is still Useful for highlighting
perceptual
necessary
and behavioural
to review briefly
speed differences.
the i,portaot
Thus, it is
C1984,
1986)
particular
Shinar,
Riemersma,
for cOllecting
free sPeed data on the road
1982).
circumstances,
needs to enSUre ContinUity of road characteristics along the road
Section of interest and should be free from any SUbstantial
roadside Obstructions.
In addition, there should also be
Sufficient length between major intersections fo, free Speed, to
be achieVed.
FOllowing identification of Potential Sites, the
actual measurement location is finally determined by safety
considerations to both operator and passing traffic.
his or her judgement of a
3.5.2 Numerical
Salvatore
have considerable
they are expensive
compound
perceptual
Measures
Speed responses
kilometres
1975; Witt and Hoyos,
While these measures
in driving,
r-
or a steering
(Denton, 1971; Rockwell, Malecki
1974; McLane and Wierwille,
validity
The criteria
stimulus event
laboratory
The subject is provided
device such as an accelerator
wheel or moves a lever to represent
aspects of cOllecting
free SPeed data 00 the rOad.
Sanderson and Corrigan
describe a method often USed by traffic engineers.
is the method of adjustment.
with a responding
responses
of_S~eed
By far the most common method of eliciting
responses
used
These include physical
and direct subjective
responses; speed limit r~ponses
of speed or safety.
3.5.1 PhYs~~al Measures
While the technique adoPted by Cairney and Croft (1985) of
matching laboratory jUdgements with free sPeed road data Was
types of measures
speed in the laboratory.
and
1976;
face
to set up and in some
and driving skill effects.
for Speed
in terms of estimated
per hour) have been collected
(1969) and Evans
(1970b).
travel speed
by Hakkinen
This approach
(miles or
(1963,
is preferable
for speed perception as it does not require a driving response.
However, subjective number continua have been shown to be nonSpeed
the rear
of measurements
an unmarked
are not warned
can be vehicle.
taken with the radar located in
parked
In this way, motorists
of the pre'ence
Parking
is not available,
the .eter
vehicle should be
However, if
should be located on
the nature strip and ca.ouflaged by adjacent trees or shrubs.
each location, 100 speed measurements of cars or their
derivatives
are recorded
(Anderson,
1970, 1981; McBride,
introduce undesirable
response
in the appropriate
At
I
Cairney and Croft
(1985) asked subjects for three variations
of speed limit responses
for each road site they presented.
These included what the speed limit should be, what their maximum
operating speed would be, and what the speed of the other traffic
would be.
traVel direction.
In addition,
they asked their subjects for jUdgements
While this was a particularly
r
.
.
- 35 --
---
in
60, 75, 90 or 100 km/h.
restrictive
- 34 -
.-'~
can
bias in speed level data.
terms of only four response categories;
--
1982) and, therefore,
of the radar and do not change
their speed.
Wherever Possible, the stationary
placed where motorists expect parked vehicles.
adjacent
linear
set of responses,
it
I
I
was also likely to evoke a sOcially acceptable
experimental
of sOmething
resPonse.
Levine
The
envirnnoent and the nature of a per'onal a.se'.ment
rather contentious in our society (the number
repreSenting the speed limit for a particular road in Victoria)
could moderate a Subject's speed jUdgement.
A more Subtle
assessment
(1977) used a similar technique
degree of safety a subject felt for a particular speed and road
environment in a simulated driving task. Watts and Quimby
(1979), too, used a mechanical
technique for assessing
derivation
SUbjective
construct and is most suitable
Furthermore, there is some doubt about a driver's ability
assess speed limits accUrately.
Moslyn and Sheppard (1980)
to
repnrted that many mntnri.t. (especi",y
oomen, ynung drivers .n'
those driving low distances annually) were not aWare of how to
reCognise
which speed limit applied
to a particular
stretch of
road.
limit
jUdgementsdata.
under these circumstances
are
not Hence,
likely speed
to yield
meaningful
3.5.3 SUbiective
Maqnitude
environment
on a driver's
of the line-mark
risk in a variety of road
situations in a vehicle simulator.
of sPeed is less likely to be biased in this manner.
to indicate the
This technique
for assessing
perception
is simple to
the effects of the
of speed.
3.6 SUMMARY
This review of the literature
and perception
on the measurement
of speed
shows that:
On-road measures
are not appropriate for sensory perception
studies as they tend to compound input effects with driving
Scalinq
skill responses.
A third approach
experiments
USes SUbjective
rate the magnitude
Subjective
for eliciting
scaling teChniques
of sensation
impression.
responses
Numbers
in perceptual
Where Subjects
have been Used for magnitude
taken to simulate
driver.
carefully
the validity
of the Power law has
been challenged (Trie.man, 1964, Anderson, 1970, 1981, MC8ride,
1982). These authors stre.s the need for magnitUde SC"e. to be
free of numbers.
A slash line resPonse was developed where
SUbjects mark across a line between two extreme values to
repre'ent
a particUlar
senSation
of an eVent.
A validation
to the
road and laboratory tests would
when setting up a new study teChnique to
any likely differences in the results between
study between
illustrate
both methods.
Moving stimulus materials
for speed estimation
dynamic retinal
presentations
displays.
than nUmber
Comparing
are preferred
over static scenes
because of the likely reliance
processes
and cues.
on
Movie film
have some minor advantage
over other dynamic
free speed data with perceptual
behavioural
. ~- •..._--._~-~
care is
be worthwhile
highlighting
-
providing
the image presented
The data u.in9
this resPnnSe
.cale1981;
has McBride,
been shoon1982).
to be less bi"ed
Scales
(Anderson,
- 36
can be useful for highlighting
aspects of driving,
adapted from techniques developed by Stevens {1961, 1962, 1974,
as
fLire~physical
measuresandofPsychological
sensation When
establishing his Power law
between
eVents.
however,
experimentation
visual perception
e.ti •• tion Of rO'd curvature (Filde., '987, Filde. and Trigg.,
1983, 1985) and cross model matching ha. been 'Pplied for 'pee,
assessment (Riemersma, 1984).
Both of these measures were
More reCently,
Laboratory
in terms of a scale of their
differences
between
consequences.
-
- 37 -
data is useful for
sensory perceptions
and
I
4.
While three methods of speed measurement in the laboratory
are available, a Subjective measure not involving numbers
appears to be the best technique available for comparing
relative effects of the environment on speed perception.
The preceding
RESEARCl:LuS'1'R1I,TEGY
introduction
cues and assessment
methods
detailed description
and literature
review of possible
for speed perception
of.a research
enabled a
strategy.
However, an estimate of travel speed may also be necessary
to compare sUbjective and objective Speed.
4.1
Speed limit responses
are not particularly desirable
dependent variables
socially acceptable because they are likely to eVoke a
response bias.
LABORATORY
EXPERIMENTATION
The literature
review suggested that speed perception
be well simulated in a relatively
simple laboratory
could
I
environment.
The degree of simulation will depend on the accuracy of the view
presented to the sUbject as well as the amount of non-visual
driving information
presented
and lateral acceleration).
experimentation,
(notably vehicle sound, vibration
As a pre-test
a validation
accuracy of the laboratory presentation
4.2
STIMULUS
to detailed
study was required to confirm the
It
method.
PRESENTATION
,;
\
Two choices were considered
materials
stimulus
to the subjects, namely static or dynamic road scenes.
While stationary presentations
laboratory
seemingly
for presenting
experimentation,
important
presentations
are simple and more controlled
for the perception
were preferred
of speed.
Hence, dynamic
for this research.
Moving road environments
can be generated
from either movie
film, video or computerised
geometric
category requires expensive
graphics and high resolution
for presentation.
inexpensive
productions.
The last
screens
video filming was the most simple and
means of reproducing
a moving road environment
was limited in terms of the size of presentation,
fine road details and difficulty
image size.
in controlling
Thus, moving road scenes were subsequently
but
resolution
automobile.
side of a car positioned
generated
using
left of the centreline of the road.
-
r-
..
of a
This means from the right-hand
- 38
- --
of
for perspective
16 mm colour films taken from close to the driving position
typical Australian
for
they do not provide the motion cues
39 -
'-"~':'.{'l".
=--------
4.3 fREE SPEED VEHICLE
DATA
4.5 CONTROLLED FACTORS
Free sPeed data Was collected
behaviour
difference •.
to highlight
perceptual and
Free .peed. were •••• Ured u.ing • e.t.o
Mini (.crn •• the rO'd) r.d.r by .0 ob.erver seated in an u".rked
vehicle at the 'ide of the road.
Thi. method wa. shown to have
least influence on the speed of vehicles being measured
(Sanderson and Corrigan, 1984, 1986).
The factors not manipulated
for experimental
remaining variables
Procedures
for collecting
with the methodology
and preSenting
recommended
levels of the
consisted
of flat, straight road sections.
the data were
by the Advisory
Co•• ittee on Road U.er Performance and Tr.ffic Code. (ACRUPTC)
Sample sizes of around 100 vehicles were adoPted.
4.4 INDEPENDENT
The following
in Table 2.3 were adopted in this research
programme.
Road alignment
compatible
in this study were held constant
control purposes.
Zero traffic density was adopted.
This meant a minimum of 5
seconds free headway in either direction was required for
both film and on-road tests.
VARIABL~
Sight distance at least twice the presentation time for each
stimulus scene was maintained.
This includes both the road
Fifteen
rOad or environment
likely to influence
factors were identified as
the .en'ory perception
Of .peed to .ome
degree (.ee Table 2.3 on page 2,). Of the.e, the effect, of the
first 5 variables Were considered here, namely;_
Urban and rUral environments,
Road category,
Lane width,
Roadside development,
Driving experience.
ahead and headway distance
same direction.
Wherever
possible,
vehicles
or pedestrians.
only had spasmodic
for any vehicle travelling
in the
sites were selected with no parked
Road tests were included
instances of parked vehicles
if they
and movement
of people.
and
All sites were filmed during daylight hours with good light
to ensure that any texture effects were accurately
in the stimulus materials.
represented
These factors
seem to eXert consistent
effects on free
'peed. on the rOad and were noted •• having been involved in rO'd
cra.he
•. Further.ore,
they con.titute a b•• e .et of v'ri'ble. 00
at
a later
time.
which the influence Of .o.e of the re•• ining f.ctor. c.n be .dded
Filming only
occurred on days of fine weather with no visible
All sites contained minimum
treatments
levels of road delineation
(centreline and edgelines were permitted,
roadside posts were removed where they interfered
filming or testing).
I
-'~"'T'"'~-~
~-'-----
signs of
rain or fog.
- 40 -
.,_,~
- 41 -
but
with
~.
4.6 PRESENTATION
SP~D
TABLE 4.1
DETAILED
The literature
review showed
EXPERIMENTAL
DESIGN
FOR THE MAIN EXPERIMENTS
that the absolute level of
speed tested was likely to influence the Subjects' jUdgements.
It is not uncommon for movement experiments on the road to
include mUltiple speed levels to minimise any chance of response
constant
estimate.
bias and speed
to ensure
that Subjects do not simply respond with a
While it would be more desirable
presentation
speeds for each site
to have a range of
(absolute speed would then
be ulD
1. RURAl.
21aneUID
9uvel
an independent variable in its own right), this manipulation
of less importance here compared with the five independent
variables.
However,
avoid guessing.
a minimum
I
,..... .....
,.I
aparce
•• l.ane
cent
of the
posted
limit.
for each
site,
and speed
were arbitrarily
U/D
.,..
SPACIOUS 1
Experi_nt
VehiCle
Travel Speed
of Melb.
• peelS
BJMllld
evaluation
,
trees
trees
.etb.clt
setback:
£HI_RURALldi.,
Inarr- ., ..
••• 7
divided
3. URBAN
a single presentation
narrtJW
•• lane UID
:2 lane
sites Were selected
4.7 DETAILED
EXPERIMENTAL
Three categories
urban) were included.
speed was held to
of environment
(rural, semi-rural and
RUral and urban environments
comprised
settings in a built up area, such as on the outskirts of a major
city with Overhead lighting and speed restrictions.
narrow
UID
speed lilllit
residen.
speed li_it
wide
residen.
:-:---
OF
SU'£S
16
--
NUMBER
or
SIT:::;
16
•
20
J
Wide ~
B
16
•
2C
3 m.
J
Ill.
Rural divided
Sparcc rural
Treed rural
Setni-rural
roads or lIrterials
roadside
refer
mainly to country freeways
develoPJllent means open country ",i thout trees
roadsicl.e develOp1llent is typically
divided
Trees for semi-rural
roads or arteriOlls
roadside
refer
..-
-~-
near highway.
D.road !>ll.ssing t.hrough a forest.
JIIll.inly to urbOln freewllys.
lIlay inclUde shnlbs ll.MJ.other
- 43 -
I
16
PRESENTI''1'lON~ •
1 &qual nufl\bers of ••ale ano feJllale first year Olndfully licenced drivers in ellch elCperi....ent;
'j ••ale elCperienced. !l felllOllc elC.pC!riencccl,!l male first
year, t1ncl!l [emOlle fIrst. year driverB.
:: Narrow ~
,
8
TO'J'~L NUHIlEJI. or DIl"FEItENT FI~
- 42 -
~
16
_ n
+/- 15 ,
+/- 15 ,
speed H.-it +/- 15 ,
residen.
wide
As it was difficult to generalise across these environments,
(rural, semi-rural and Urban) was constrained to a seperate
. :=....:... ~.. _. --
NUMBER
wide
experiment.
The variable combinations and factor levels used in
each experiment are described fUlly in Table 4.1.
r
16
DESIGN
country and city settings and were relatively clear cut. Semirural environments, on the other hand, were defined as rural
each
liJIit +j- 15 ,
+/- 15 ~
+/- is ~
speed li_it +/- is ,
speed H_it +/- is ,
1'·<hOck
'tOTAL
the
SpeedSpeed
limit.
with posted
differing
limits and presentation
16
15 ,
speed l1alt
:2 lane UID narrow wide
lane UID
vide
study,
+/- 15 ,
liait +/-
'I'O'l'AL N\JIllIeR OF SITES
'I'O'l'AL
For the laboratory
rr •• entationa
12 for •• eh
J
speed It..-it
, ••
speed was used because of the road trials,
Road Sit ••
tall within
:2hr.. of
drive INUllber
type ofofsit.e1
Nu.l:MIr
w.u.un
speed. liait.
15 per
0(
Roadside DevelOt-ent
was
by the speed limit
set at plus and minus
width
~e
tree.
tree
• aparce
narrow
n&1'1'Olof
wide
vide
divlded2
aparce
aparce
narrow
J wide
of two speeds was necessary to
These speeds were determined
....• ,.,...
&xperi_nt
low lying foliage.
'0
"
FIGURE 4.0
4.8 STIMULUS MATERIALS
Research Vehicle with Camera in position
The stimulus materials
for each site consisted of a 5 second
sequence of 16 mm film, taken from the driving position of an
automobile
left of the centreline.
Figure 4.0 shows the camera
\
in place on the experimental vehicle.
A 10 second section of
blank film was inserted between each test scene to allow the
subject
sufficient
time to respond in a semi-paced
A pseudo-random
separate
of stimulus
within
order of presentation
rolls of film, each containing
scenes, were prepared.
each film was consistent
of presentation
\
manner.
was adopted, where
a discrete
fixed sequence
Thus, the presentation
for all subjects,
of each film was systematically
and instructions
order
but the order
varied between
subjects.
Practice materials
preceded the main
experiment
to ensure the subjects were familar with the task.
4.9 SUBJECTS
The subject pool consisted
drivers
of equal numbers of first year
and full license holders.
level of driving
experiments,
experience
yielding
Eighteen
were recruited
sUbjects of each
for each of the 3
a total of 108 subjects.
The subjects were drawn from within RACV Limited and from
specific
populations
possible,
of young drivers
sUbject pools were matched
experience
and exposure.
participating
Subjects'
outside
RACV.
As far as
for sex, years of driving
In all cases, sUbjects were paid for
in the experiment.
vision was screened
at the start of the experiment
to ensure they had normal vision
(1.0 Snellen or better).
included
spectacles
driving
people with and without
population.
- 44 -
r-
This
to match the normal
""'~-'~~""',,-""~'~'.~.
",.,,,,,,,"'-,,,.
~==:::::::::::;;;;;;=-'i.,....---=.,....-=._~_._---,
- 45 -
4.10 DEPENDENT
FIGURE 4.1
VARIABLE
SAMPLE PAGE FROM THE RESPONSE BOOKLET
The laboratory
only a visual
method adopted
simulation
response measure
variable here.
in this research comprised
of the driving task.
was not an appropriate
Thus, a vehicle
or desirable
dependent
W
E-<'
o
E-<
tn:tl
objections
However,
it
estimate
response
effectively
to using speed limit responses
is essentially
a relative
overcomes
E-<
o
8~
WH
P<E-<
A magnitude
1
I
......-
tn
-r
'"
p..
W
for this task.
speed measure and may not
provide data that can be directly compared with the free speed
data from the road.
Thus, subjects were asked to make TWO
responses to each road scene presented.
The first response was a
slash across a 150 mm line, labelled at each end as either too
fast or too slow. A mid-point was also provided but not
labelled.
The second response was an estimate of what speed the
subject believes he or she is travelling at. These responses
were used to compare with the free speed data collected at each
site.
Figure 4.1 shows a sample page from the response booklet
used in the experiment.
Each jUdgement was on a separate page to
minimise carry-over effects from one jUdgement to the next.
4.11 DATA ANALYSIS
Statistical
analysis
of the experimental
data involved
parametric statistics.
Keppel (1982) noted that the most robust
test of experimental data is analysis of variance.
He also
0::
r.u
co
:<:
:::>
z
r.u
recommended the use of the omega-squared statistic for assessing
the relative strength of effect of each of the variables.
Both
of these omnibus tests of significance were used for data
analysis, along with appropriate
necessary.
H
UJ
oo
E-<
.L:::
o
....:I
tn
simple effects analysis where
- 46 -
-----------".--- ----.---_-o---"-:-:~
r-
E-<
- 47 -
These techniques
precise
form of analysis
available
for these experimental
Free speed data were summarised
analysis
package and presented
percentile
5.
allow the full effects of each experimental
variable to be measured, as well as any interactions that may
occur between the variables.
This is considered to be the most
and standard
data.
using a standard free speed
in the form of mean, 85th
deviation
for each variable
combination.
could not be collected
at all sites
A preliminary
LABORATORY
VALIDATION
study was undertaken
STUDY
to validate
the method
proposed of measuring speed perception in the laboratory.
As
well as helping to establish a valid experimental procedure, this
'study would also provide valuable
drivers' perceptions of speed.
insight and practice
in testing
5.1 STIMULUS MATERIALS
Free speed measurements
Twelve urban road sites in the Melbourne
Metropolitan
area
for a number of reasons (eg: sites were by-passed between filming
and free speed measurement).
In addition, there were variations
in the numbers of vehicles at some sites due to the lack of
were selected as typical examples of a range of road sites to be
traffic movements
5.1 for complete details of the sites selected
analysis
and time of the year.
of these data therefore
a descriptive
here.
analysis
A full statistical
was not possible
and hence only
of the free speed responses was attempted
used in the main experiment
(see Appendix
A-1 and Figures
5.0 and
for the validation
study).
These sites offered a variety of road categories, speed
limits and road environments and in all cases, there was a
minimum
of 30 seconds unobstructed
sight distance.
A 5 second road segment was selected
study and free speed measurements
at each site for the
were made at each of these
12
experimental segments (details are listed in Appendix A-1).
A
film sequence at the posted speed limit was taken at each site
and sUbsequently spliced into a 5 second segment as described
earlier.
This film sequence was later used for the laboratory
assessment.
Road trials were conducted
at the same road site.
5.2 ROAD TRIALS
Six fully licensed drivers
(3 male and 3 female) with no
prior knowledge of the project were recruited from within RACV
staff and driven around a pre-determined course encompassing the
12 chosen road sites
(see Figure
5.1).
Each subject sat in the
passenger
side front seat and was instructed
procedure
from a cassette tape, prior to arrival at the first
site (Appendix B-1 lists the experimental
the road experiment).
- 48 -
-
49
-
in the experimental
instructions
used for
::P.
tu
0.
n
tu ~ ::l
c'"'£'
0'
tu
~ I ~~,
"3
.':4;'~<;:~~:::~;
'£' t'~~:i;
r
c[
::l
0.
(fl
0:.
<'
(t)
t:u:l
-0
'-l
tu
-l>ero
-~
,"
00
~
oz
C::V1
)-m
<~
-<c
;:oC')
O:!!
~
o;:0
OJ
r)-
As each site approached,
adjusted
the driver of the experimental
his speed to the posted speed limit and positioned
vehicle to maximise
free headway
headway was adopted as a minimum
car
the
(as a rule, a 5 second free
trial requirement,
although no
00
traffic was always aimed for at each site).
w
The subject was
'"
o
•...
asked to look down at the response
and to look up only when instructed
the back seat.
shielded
The speedometer
from the subject's
through the vehicle's
noises.
As the vehicle
w
booklet on his or her knees
by the experimenter
of the experimental
vehicle was
entered
each experiment~l
site, the subject
time, the subject was then
After 5 seconds of viewing
asked to look down again and make his or her assessments
safety and estimated
oW
0:1:
"''''
[8]00
'"
travel
speed.
After
ll]J
~
lZ1
:2
u.J
l..l
Z
10 seconds, the
was given before the first experimental
the subject with the task.
number of subjects
involved
for the experimental
'"
'"
o
trials
for the laboratory
conditions
trial to
Because of the minimum
in this study, a fixed route was used
(a fixed presentation
presentation
order was also
of the 12 movie film
All road trials were conducted
5.3 LABORATORY
==
~
«
:><
00 •..•
Z
o
during Off-peak
and with dry roads and good light conditions.
lUJ
>,~
TRIALS
'"
used for off-road
of the road trial method.
•...
Z
"'w
II II
1\
C1.
)(
I I
W
I I
I \
I
I
I I
-'
w
wz
~~
and natural
of the laboratory
of the RACV in a
light and only had one access door).
arrangement
zzw
~::: I
\
,
:l:C1.'"
II
I
~~~i
-0-'
\
II
w
I
__
\
l:u_--
I
__
__-
-'<
~:I:
\
(/)c:
--- ---\-....
,
l-----·~
- 53 -
_---~---
w
-,
t
UJ
:
w
N
.....
~ ~I
i3
t::
If)
LU
0::
@)~~
t; ;~\
:; 0
"
_
Details
._._ .. _ ..
:>000
-'wo
=a::m
<eO
...Ja::
\
are shown in Figure 5.2.
- 52 -
~ : 4:
lw
•...
w
<00 •...
I
I
\~\~
\::1
ow
I
,
(the room was without
CO
II
:I:
A laboratory was
quiet works area
.«
11
w
testing was a close
at the Noble Park Headquarters
and relatively
~
.~
o
1\
I1
w
z~\
The procedure
:0
01
~
oot;
•...w
windows
CID Z«~
-'
•...
-,ID
u..w
isolated
lJ
zo
~
<w
W
ID
l..l
Practice
established
Z
u.J
:l:
of
and traffic.
replication
I-
CID
experimenter instructed the subject to relax and engaged him or
her in casual conversation to distract attention from the road
traffic
[j"
:1: •...
iii
otherwise.
segments).
l..l
z
ahead until instructed
adopted
-'
<
u;~
view and loud white noise was played
was asked to look up and view straight
familiarise
w
'"
>wlD
NOO
seated in
stereo system to mask engine and wind
~
:I:
IJ,..
An experimental
constructed,
along with another
sites for practice.
recruited
tested
film sequence
The speed estimate responses were SUbsequentlY
of the 12 road sites was
film of 6 additional and novel
Six new subjects
(3 male and 3 female) were
from within RAC V headquarters
and were individually
in the laboratory.
converted
error scores (positive or negative differences in km/h
into
estimate and the actual travel speed) and also
1 of
between the
analysis of variance and omega-squared.
The
analysed using
of these results is also shown in Table
statistical summary
APpendiX C-1.
Each subject
experimental
practice
listened
instructions
to a tape recording
(see Appendix
describing
B-2 for details).
the
5.4.1 ~fe
The
operatina
Speed Data
film was then loaded and the subject was given practice
at making
responses.
experimental
The practice
sequence
speeds to prevent
The main effect for road site shown in Figure 5.3b was
film was identical with the
but used different
significant
sites and different
(F(11,110)=9.7,
considerable differences
learning.
12 road sites used in this experiment.
The experimental
film was then run.
manipulation
Each road site was
presented in the same manner as the road trials.
Subjects viewed
the road scene until the onset of the 10 second blank, then
looked down and made their assessments
travel
experiment,
treatment variance
of the experiment,
about their impressions
This variable
effect observed
for almost one-third
in this
of the total
in this experiment.
There was also a significant
analysis,
At the conclusion
expressed
of safety across the
of safety and estimated
speed as previously.
questioned
was the strongest
accounting
subjects
p<.001, w2=.292).
in their judgements
shown in Figure
main effect observed
5.3a (F(1,10)=21.1
, p<.001,
in this
w2=.165).
The mean estimates of safety in the laboratory were approximately
subjects were
equal to the mean scale value, whereas the road estimates were
significantlY slower.
This variable accounted for approximatelY
17 per cent of the experimental variance, but even 50, was only
of the task and strategies
they used when making their estimates.
one-half the strength of statistical
5.4 RESULTS
association
as that of road
site.
The safe operating
speed responses
for each site in both the
A significant
interaction
road and laboratory experiment were measured and converted to a
distance in millimetres from the left hand "Too Slow" anchor
experiment
point.
2.3, p<.05, w2=.042).
Thus, any number between 0 and 75 represented
response
while numbers
between
a slow
and site variables,
was practical
75 and 150 indicated a fast
was observed between the
shown in Figure 5.3c (F(11,1101=
The pattern of responses
occurred in the results for site 2.
The response
and analysed
numbers were then structured
using the ANOVALEE
and omega-squared
statistics
program
12 values of a within-subject
of variance
into a data base
for analysis of variance
(Fildes, 1987).
and 2 values of a between-subject
analysis
(albeit at
a higher overall level of safety) except for the reversal
response.
assumed
in the laboratory
identical to that of the road responses
variable
variable
The analysis
for "road site"
for "experiment".
summary tables are shown in Appendix
The
C-1.
- 55 -
- 54 -
that
5.4.2 SpeedE~timation
A significant
(a)
(b)
SS
z
7sr·---'\.
95
w
site tested.
./'-"
:lE
"
w
1..""-. __
Q
:)
..,
type of experiment
W
D.
SLOW
Cl)
'---ctB·
Z
j:
R~D
TOO
SLOW
TYPE OF E-XPERIMENT
TOO
L1
,
,
,
,
,
,
•
a:
(c)
,
,
'-,
1 2 3 4 5 6 7 8 9 10 11 12
TEST SITES
Cl: FAST
effect
captur~ng almost eight per cent of the
interaction
observed between
and site shown in Figure 5.4c (F(11,110l=2.3,
p<.05, wZ=.005l.
A simple main effects analysis performed on
these data shows that the only significant differences were at
TOO
"
This variaple was again the strongest
observed in this analysis,
total variance.
There was also a significant
Q
W
main effect for site was observed as shown in
Figure 5.4b (F(11,110l=2.B, p<.01,
wZ=.007l. The accuracy of
the subjects' speed estimates was dependent on the particular
Fl~~1
'-./
••••
Data
EXPERIMENT X SITE
sites 5 and 7 (F(1,10l=6.6, p<.05 and 9.B, p<.01, respectivelYl.
None of the other sites were significant
at the 5 per cent level
of probability.
W
D.
o
There was no significant
W
IL
.-.
Cl:
1'''., _.
Lab Sale ~im~\.
·""x
.....
'\,,,.--0."
"',- / I1' __ /
Cl)
55
'--".---/Road
"'_;0-
Safe Estimate
analysis,
wZ
in this
= Ol.
Subjects' speed estimates were not noticeably d~fferent in the
laboratory to those made during the road experiment, even though
the trend was similar to that observed for the safe operating
speed data (Figure 5.3al.
TOO
main effect for experiment
shown in Figure 5.4a (F(1,10l=<1, p>.25,
Moreover,
the lack of a significant
main effect for site for either laboratory
SLOW
Lt
I
I
t
I
I
I
I
;
f
I-L
1 2 3 4 5 6 7 8 9 10 11 12
TEST SITES
confirms the uniformity
speed estimates.
or road testing
in the response pattern for both types of
5.4.3 Free Speed Data
FIGURE 5.3 Safe operating speed responses obtained in
the laboratory validation study.
Free speed data was collected at B of the test sites and is
summarised
in Appendix B-1 with the site definition
In addition,
information.
the mean speed values observed at each site, along
with the posted speed limits, are plotted in Figure 5.5.
reasons previously
these data.
explained,
For
no formal analysis was performed on
Hence, the report of these results will be confined
to a descriptive
analysis only.
- 56 -
57
-
-------.
I
,
FASTER
(a)
FASTER
+10
(b)
+ 10
.:
•...
E
:.
a:
o
a:
FREE SPEED X LIMIT
ZERO
ZERO
. ----------10
-10' ./'
a:
III
III
t-
L
I
1
LAB
ROAD
TYPE OF EXPERIMENT
FASTER
Q
,
2
,
3
I
4
I
5
I
6
I
I
7 8
I
I
•
/'\
a.
/
Cl)
ZERO
•
I/\
\\
I
"",
::
-'-"'-'Free
.5peed 5pe
LI~~ls (mean)
Do
Cl) 40
TEST SITES
III
III
a:
30
IL. 20
10
/'\ '.
L
\.
o U
51
~'I\
~
X•.
I
'0_
III 50
I
9 10 11 12
Lab Speed Estimate
+10
-10
I
1
(c)EXPERIMENT X SITE
III
,
III
SLOWER-l
~
'" .
0'60
E 70
:lE
III
III
-= ...
t
100r_\~~
\.
~
SLOWER
ee
t:
en
90
/ j"- /.-",'.'1'-·'-. .
I
52
I
53
I
54
t
55
I
56
I
57
J
511
SITE NUMBER
/'>\l:; .-.
\
SLOWER -"
1
V
,
2
/
.-
,
3
,
4
~ad
,
5
,
6
,
7
Speed Estimate
,
8
FIGURE 5.5 Free speed measurements in kmlh taken
at 8 of the 12 test sites in the laboratory
validation study.
, , , ,
9 10 11 12
TEST SITE
FIGURE 5.4 Errors in the travel speed estimate results
from the laboratory validation study.
l
i
I
- 58 -
r:--'~
_=---~~~=~=====~.
::===::.:. -:-.;:.",::;.';;;;.",~"!;~!'~~-::!E:-==::-;:a>!'''!''!''...,.,.,,==----
------
- 59 -
65
85
(b)
(a)
The mean travel
speed was approximately
speed limit at sites
1, 2, 3 and 11.
speed limits ranging
from 60 to 100 km/h.
-~
equal 85to the posted
These sites encompassed
FAST
TOOl
65
75
I
F~~~\
75l
However,
the mean
travel speeds at sites 4, 5, 6 and 7 were all approximately
10
km/h faster than the 60 km/h posted speed limit.
In addition,
sites was between
limit, depending
environment.
&
5.4.4 Road
the 85th percentile
on the posted speed level and type of roadside
jUdgements.
variables
variable
Z
III
TOO
SLOW
of the validation
A restricted
combinations
evaluation
study was to test the
for eliciting
road speed
range of road and environmental
to the main experiments
were not exhaustive.
Nevertheless,
and the
a
of these effects was still possible,
providing
care was taken not to interpret
too much from these
results.
estimates
Figure 5.6 shows the plots of the safe operating speed
for the road and environment factors included.
Several
:)
.,
DU)
"Z-
assessed
to be a safe operating
slower than divided
et
a:
III
65
D-
O
III
~
65
TOO
SLOW
s;:g~
DIVIDED
UNDIVIDED
ARTERIAL
(e)
FAST
85
.-
roads were generally
roads, while for divided roads, 8-
75
Thi~ result was independent
of whether the
65
road was a major arterial or a freeway.
On undivided roads, 4lanes were assessed much slower than 2-lanes
The main reason
for this, strangely enough, appears to be 4-lane collector and
local roads, rather than similar arterials.
TOOL
SLOW
I
'
'
Residential SemioruralResidential
LSPACIOUSJ
'
Commercial
LWALLED INJ
ROADSIDE ENVIRONMENT
FIGURE 5.6 Safe operating speed judgements for a range
of road factors studied in the laboratory
validation study.
- 61 -
- 60 -
--.--=--------------_ ..~~-----~
COLL/LOCAL
ROAD TYPE (4-Lane Undivided)
ROAD TYPE
lanes were jUdged slower than 6-lanes and 6-lanes were judged
slower than 4-lanes.
75
to-
speed when the data was
Undivided
85
----.
75
TOO
across experiment.
(d)
TOO
FAST
et
First, all roads were judged slower than that normally
collapsed
2·L
NUMBER OF LANES (undivided)
85
Q
1&1
III
U)
considered
4-L
4-L
(c)
TOO
FAST
U.
points can be made from these figures.
6·L
NUMBER OF LANES (Divided)
Q
of the laboratory
8.L
"
Effects
was used here compared
preliminary
TOO
SLOW
to-
:i
III
Environment
The main purpose
suitability
speed at each of these 12
5 and 38 per cent greater than the posted speed
'-
Walled roadside
environments
above a safe operating
which were assessed
in walled
speed, compared
generally
residential
resulted
to spacious environments
as too slow.
environments
The free speed data showed that speed l~~~ros'were generallY
in judgements at or
In particular,
appeared
speed
much too fast on
not adhered to at the test sites, especiallY
limits.
those with 60 km/h
The average speed for these lower limit sites was
approximatelY
10 km/h above the posted limit and B5th percentile
average when viewed at the posted speed limit.
speeds were up to 22 km/h (37 per cent) faster than the legal
level. This result needs further testing in both rural, semi-
5.5 DISCUSSION
rural, and urban environments and has important
the role of perception in driving.
The results of the validation
appropriateness
simulation
While the safe operating
for eliciting
road
speed average was
operating
speed jUdgements
generally
observed
and speed estimates.
from laboratory
consistent
across
method
in each condition
laboratory
significant
trials),
of road and environment
without placing too much
and
emphasis on these results, they nevertheless suggest that a
travel
Systematic evaluation of these variables will uncover new
important reasons for understanding
at the speeds they do.
why drivers choose to
The interaction
with onlY a small number of
(n=6 for both the road and the
results obtained
confirm
pattern of results
speed
testing.
judgements
observed
the safe operating
slightly
more sensitive
in the results
and speed estimates.
from both these measures
although
and the highly
that the effects are robust
There was not a lot of difference
safe operating
measures
speed estimates
to the independent
The
was quite similar,
may have been
variable
manipulations.
It should be noted though that the experimental
not orthogonal
(each variable
of the other variables).
was not represented
Moreover,
a single example of a particular
is still necessary
to collect
travel speed judgements
design was
at each level
in some cases, there was only
variable
combination.
Thus, it
both safe speed estimates and
in any future testing.
- 63 - 62 -
~
factors
findings.
and test sites was minimal
the use of repeated
and negate the need for additional
between
examination
some interesting
from only one of the sites used in this study.
While this result was achieved
subjects
has been
trials and measures.
between presentation
and resulted
In other words,
presentation
The preliminary
highlighted
higher for laboratory presentations than road trials, the pattern
of results obtained was similar at each test site for both safe
the loss of reality
for
study confirm the
of using laboratory
speed perceptions.
ramifications
,
6.
THE RURAL_EXPERIMENT
The pilot study confirmed that the laboratory
method
elicited similar speed responses to those taken on the road.
addition, the preliminary
substantial variations
speed estimates.
in the safe operating
Thus, the laboratory
to be a suitable means of evaluating
changes on drivers'
single experiment
replicated
the variables
and urban
in each of these environments
needed.
Hence, separate
environments.
for each of these environments
separately.
However,
draw general conclusions
and a
in terms of the
were designed for the three different
The experiments
described
and
seemed
and their levels
would have been unmanageable
number of stimulus presentations
experiments
method
the effects of perceptual
for rural, semi-rural
was not possible;
were not perfectly
speed responses
presentation
speed estimates.
A single experiment
environments
In
site data from the pilot study showed
will be
the final chapter will attempt to
that can be made for all environments.
6.1 STIMULUS MATERIALS
In line with the research strategy outlined
twenty-eight
Melbourne
which encompassed
particular
in chapter 4,
rural road sites were located within 2 hours of
the range and levels of the
road and environment
factors of interest.
Appendices
A-2 to A-5 detail the 28 sites selected and Figure 6.0 shows
typical sites used in the rural experiment.
The independent
divided-wide
wide
(D/W),
variables
(2L/W), 2-lane-narrow
narrow
(GIN),
repetitions
included seven road types,
divided-narrow
(DIN),
4-lane
(2L/N), gravel-wide
two roadside environments
(4-L), 2-lane-
(G/W),
gravel-
(spacious, walled),
two
(different sites with the same characteristics)
and
two film speeds
(15 per cent above and below the posted limit).
This yielded 56 road scenes in a fully crossed factorial design
experiment.
--,
-- ------::
-
65
-
A suitable road segment was identified
FIGURE 6.0
RURAL
ensured a minimum
SITES
at each site that
of 5 sec film time with specified
distance requirements.
sight
Each site was then filmed from the
research vehicle providing the two film segments,
15 per cent
~bove and below the posted speed limit.
Sites were filmed using
a Bolex H16 reflex movie camera on 16mm Kodak colour negative
daylight film (64 ASAl, processed
subsequently
into work prints and
edited into 5 sec sequences.
Four experimental
films were then produced,
each containing
selected road scenes followed by 10 sec of blank
Care was taken to ensure that road scenes with similar
14 randomly
film.
characteristics
(repetition sites or the same site at different
speeds) were on different
reels or were spaced as far apart as
possible.
The order of presentation
structured
across subjects to ensure that each film was presented
of the four reels was
equally in every serial position.
Site 9 - Undivided.
4-Lane, Spacious
road.
6.2 EXPERIMENTAL
Thirty-six
the experiment.
groups
PROCEDURE
licensed drivers were recruited as sUbjects
Nine sUbjects were allocated
(male-experienced,
female-experienced,
f,"male-1st yearl to test for experience
perception.
male-1st
participation.
in a session
40 min and were paid for their
The laboratory
used in the validation
year,
or sex effects in speed
SUbjects were tested individually
lasting approximately
for
to each of four
study
set up was identical with that
(see Figure 5.3 in the previous
Chapterl.
Each subject was played a pre-recorded
experimental
instructions
appendix B-2 for detailsl.
.~
was then presented
requirements
at the start of the session
for practice.
Questions
about task
were answered prior to commencing the main
2-Lane, Gravel wide, Walled road.
- 66 - 67 -
the
(see
The film used in the previous
experiment.
Site 24 - Undivided,
tape describing
study
~/
(b)
(a)
A
The experimental
procedure
TOO'
95
FAST
__
was similar to that used
•.......••.....•...
75
FAST
previously.
For each experimental
trial, subjects 95viewed the
75\-- The subject
moving scene for 5 sec until the screen went blank.
then looked down at the response
assessment,
repeated
followed
by an estimate
until the film ran out.
of travel speed.
trials were completed.
experimenter
highlight
discussed
strategies
This was
SUbjects were encouraged
rest while the film was changed and any questions
purpose of the experiment
mo~
book and made his or her safety
were deferred
to
relating to the
l-
At the end of the session, the
the experiment
or difficulties
55
55
until the experimental
with the subject to
associated
with the task.
Z
u.I
~
TOOL
SLOW
l'DIW DIN, 4L, 2UW, 2UN
, G/W
' GIN
,
TOO
SLOW
SLOW
FAST
PRESENTATION
SPEED
TYPE OF ROAD
u.I
o
Q
~
6.3 RESULTS
.,
As before,
the safe operating
judgement were converted
and 150 = too fast.
estimates,
into millimetres
These responses,
were computed,
of variance
speed responses
these two analyses
will be described
standard deviation
and 85th percentile
measurements
were also computed
between 0 = too slow
and analysed using analysis
statistics.
For ease of reporting,
separately.
D-
UI
o
-l-
Z
Cl:
a:
The mean,
75
75
55
55
u.I
D-
of the free speed
and reported
95
u.I
u.I
along with the speed error
unscrambled
and omega-squared
95
Q
for each
in a later section.
(d)
TOO
FAST
(c)
TOO
FAST
O
TOO
SLOW'
TOO
SLOW
u.I
SPACIOUS
u..
ROADSIDE
FEMALE
WALLED
DRIVER
ENVIRONMENT
Cl:
6.3.1 Sa~e Operatinq
UI
Speed Responses
Appendix
2-way interaction
described
95
C-2 lists the summary table of the analysis of
variance of the safe operating speed data while Figure 6.1 shows
the effects of interest.
There were four main effects and one
that warrant
close attention
in order of their strength
and these are
of effect
(l)SPEED X ROADSIDE
TOO
FAST
(e)
TOO
FAST
MALE
SEX
95
·Spacious
75
75
ranking) .
55
55
Presentation
effect
speed shown in Figure 6.1b was a significant
(F(1,32)=118.3,
pC.001, w2 =.0895).
Subjects estimated
slow speeds to be too slow while fast travel speeds were judged
too fast.
This variable
was the strongest
effect recorded
=.0895) and captured more than half the variance
independent
variable
manipulations.
- 68 -
(W2
TOOl.SLOW "
TOO,SLOW l'
EXPERIENCED
DRIVING
'
NOVICE
EXPERIENCE
due to the
FIGURE
6.1 Safe operating speed results of 'Interest
for rural roads in exper"lment 1.
- 69 -
SLOW
PRESENTATION
'
FAST
SPEED
(b)
Type of road in Figure
18.7, p<.001,
w2=.0484).
6.1a was also significant
Subjects
(F(6,192)=
estimates varied from too slow
+~
for major arterial roads through to too fast for minor and gravel
roads.
This variable was the second strongest effect ( w2=.0484)
and accounted
Roadside
for almost
environment
(F(1,32)=40.2,
of spacious
variable
p<.001, w2 =.0165).
film segments
were assessed
effect as presentation
too fast.
l"
ZERO
/.--
to be safe, whereas walled
With
w2=.0165, this
only one-fifth as strong an
---.
shown in Figure 6.1d almost reached
(F(1,32)=3.9,
was a between-subject
p=.055,
w2
,.,,,, ZERO
./".
.---------- .
----.
- 20
I
-
"
SLOWER
,
DIW
DIN
.c
,
,
,
,
,
4L 2L1W2L/N GIW GIN
SLOWER
effect.
This variable
accounted
presentation
for 5 per cent of the variance
the strength
of effect as
~
a:
oa:
a:
+ 20
+20
ZERO
c:[
~
t/)
-20
-20
L&I
Q
L&I
between
speed and roadside shown in Figure
SLOWER-I'
sites were judged slower than walled
sites
p<.001), while for fast presentations,
much faster than spacious
however,
sites
walled sites were assessed
was not a particularly
only 1 per cent of the variance
treatment
manipulations.
ROADSIDE
t/)
SLOWER
MALE
ENVIRONMENT
(e)
FASTER
(F(1,32) = 38.7,
(F(1,32) = 65.1, p<.001).
capturing
'
WALLED
SPACIOUS
L&I
Do
6.1f was also significant (F(1,32)=6.3, p<.05, w~.0014).
Simple
effects analysis showed that for slow presentations, spacious
interaction,
(d)
FASTER
ZERO
•••
•••
speed.
The interaction
+20
This
strong effect,
attributed
to the
ZERO
-20
6.3.2 Speed Estimation
Errors
To provide a meaningful
travel speed estimates,
scores by subtracting
Appendix
FAST
SPEED
L&I
factor,
data here could result in sex being a significant
and was less than one-tenth
-~---
SLOW
PRESENTATION
FASTER
(c)
L&I
additional
_~,
TYPE OF ROAD
E
at the 5 per cent level
As this variable
_ 20 L
estimated the speeds
speed.
The sex of the subject
=.007).
Subjects
was ranked third, although
significance
""'''I
30 per cent of the treatment variance .
shown in Figure 6.1c was also significant
environment
environments
+~
and comparable
analysis of the
the raw data were converted
SLOWER
••..
L
----',-----EXPERIENCED
DRIVING
into error
NOVICE
EXPERIENCE
the actual speed level from each estimate.
C-3 lists the analysis
of variance
these error scores while Figures
summary table of
FIGURE
6.2 Travel speed errors for the main effects of
interest for rural roads 'm experiment
1,
6.2 and 6.3 show the effects of
- 71 -
70
-
FEMALE
DRIVER
SEX
"
I
r
I '
r
.
J
f'
prime interest here.
Three main effects and four interactions
are described in order of the strength of effect.
Presentation
.I
FASTER
-
(a)ROADS X ROADSIDE
I
(F(1,32)=285.4,
FASTER
(b)SEX X SPEED
+20
ZERO
a:
a:
a:
-20
~
o
•
•.........•
~ 7·~led us
/,
__ ••
/
._===-.~__
."","
;;-._. Spacio
Type of road in Figure 6.3a was significant
et
SLOWER
-L
'£
;:
en
f
,
-20
,
I
f
I
SLOWER
FASTER
"~'Male
"
under-estimated
.• Female
FAST
PRESENTATION
p<.001,
accounting
also
In general terms, subjects
w2=.1041).
the speed of major arterial
estimates became more accurate
This variable
SLOW
(c)SEX X ROADS
(F(16,192)=59.4,
'-----'.
DIN
w
Q
W
W
D.
en
,
4L 2l1W 2l1N G/W GIN
TYPE OF ROAD
D/W
.~
ZERO
w
w
l-
The speed of slow
w2=.1398).
strongest effect observed, capturing 50 per cent of the variance
attributed to the variable manipulations.
...
E
p<.001,
presentations was judged accurately, while fast presentation
speeds were under-estimated.
Once again, this variable was the
+20
.c:
speed, shown in Figure 6.2b, was significant
roads while their
as the road category
was the second strongest
reduced.
effect observed
again,
for 37 per cent of the factor variance.
SPEED
FASTER
(dISPEED X ROADS
An interaction
+20
was observed
roadside environment,
+ 20
between type of road and
shown in Figure 6.3a (F(6,192)=15.9,
p<.001, w2=.0146).
Simple main effects analysis confirms that
the source of this interaction was for divided wide
ZERO
•
.._,
-20
__ •
...•.•.•......•
__
e__
Female
/---x:-,.'" ---Male
./
ZERO
_'Slow
.",.--.-./
/
-20
•••......
SLOWER
L
I
D/W
I
,
DIN
I
I
1
r
4L 2l1W 2l1N G/W GIN
--.-
---_/
/.-
(F(1,32)=9.05,
lane narrow
•.•.•• Fast
_.-
(F(1,32)=35.0,
SLOWER
TYPE OF ROAD
~
D/W
p<.001).
narrow
p<.01), two-
None of the other road categories
at the 1 per cent level.
interaction
accounted
With w2=.0146,
were
this
for 5 per cent of the treatment
variance.
"---,
DIN
4L 2l1W 2l1N G/W GIN
TYPE OF ROAD
There was also a significant
subject's
interactions
6.3 Travel speed errors for the significant
observed for rural roads in experiment
1.
significant
w2=
.0107).
and 22.7, p<.001 respectively)
I
_I
road type
the
for 4 per cent of the treatment
analysis.
I
- 73 -
showed a
and female subjects
but no significant
(F(1,32)=2.0,
3.8, 2.5, 0.6 and 0.4, p<.05 respectively).
accounted
between
Simple effects analyses
type of road main effect for ~ale
(F(6,192)=16.9
I
I
interaction
sex and type of road, shown in Figure 6.3c (F(6,192)=
sex effect for any particular
- 72 -
(F(1,32=8.4,
p<.001) and gravel narow roads
significant
7.0, pC.001,
FIGURE
p<.01), divided
(F(1,32)=133.2,
2.6, 3.4,
This interaction
variance
in this
The grand mean free speed across all measured
The interaction
between sex of the subject and presentation
speed, shown in Figure 6.3, was significant (F(1,32)=8.7, P<.01,
w2=.0038)
Simple effects analysis, however, did not reveal
any further information;
speed was significant
for both male and
female drivers (F(1,32)=22.1 and 44.3, P<.001 respectively), but
sex was not significant for either slow or fast presentations
(F(1,32)=0.5 and 2.7, P>.05 respectively).
This interaction
accounted for only 1 per cent of the treatment variance.
than the mean posted speed limit
km/h).
sites was less
(96.7 km/h compared with 100
The mean standard deviation
for these sites was 11.4
giving a mean 85th percentile value of 108.6 km/h or B.6 km/h
above the posted speed limit.
It is unclear whether these values
are statistically
different
from the posted speed, given the
uneven nature of the sites measured
standard deviations obtained.
and the relatively
large
Figure 6.4 shows the plots of the mean free speed deviations
The interaction
between
presentation
Speed and tYPe of road,
around the speed limit for the available
roadside environment
shown in Figure 6.3d, was also significant (F(6,192)=5.5, P<.001,
W2=.0025).
Simple effects analysis revealed a significant main
effect for type of road at both slow and fast presentation speeds
(F(6,192)=36.6 and 53.4, P<.001 respectively) as well as a
These figures
significant
free speed seemed to reduce as the type of road and number of
speed effect
for each road type
(F(1,32)=163.2,
145.5, 118.8, 169.9, 86.7, 119.5 and 53.2, P<.001 respectively).
This interaction captured less than 1 per cent of the total
treatment variance in this experiment.
also include a plot of the 85th percentile
show that walled environments
lanes reduced
effect
(omega-squared
(F(1,32)=
However, sex was a between-subject factor
and did have a noteworthy strength of
values greater
than
.001 deserve detailed
attention; Triggs, Harris and Fildes, 1979; Fildes, 1979, 1987;
and need to be tested more stringently in any future
experimentation,
Keppel, 1982).
comparing
the effects of type of road with
and divided roads had mean free speeds around the posted speed
limit, while
2-lane
(and especially
2-lane-wide
bypassed between filming and collection of speed measurements,
while for the gravel sites, only site 22 had sufficient traffic
to enable free speed measurement).
The mean speed, 85th
environment
influences
With the exception
of roadside environment
of 2-lane-narrow
these values were consistently
- 75 ":':=-
- ..
-::::::-,
'::::.:::::::::::::::=::;.,===~.......",
~d :::::.. :::;::;::,.
_
and type
roads, however,
above the posted speed limit
maximum value of 113 km/h was recorded
sites) .
- 74 -
. dO,:=:"" ..2=.~"-'=-"""
free speeds on the road.
plots show similar trends to the mean
percentile and standard deviation of the free speed records taken
at each site are listed in Appendices A-2 to A-5
.
12.5 km/h
results for
roads suggests there may be some optimum value at
which roadside
of road.
Free speed data were collected at 20 of the 28 selected road
(at site number 10, the road was realigned and the site
narrow roads) appeared
noticeably less than the speed limit (approximately
below).
The difference between spacious and walled
plots in terms of the influence
sites
to result in
(Figure 6.4a) while
shown in Figure 6.4c suggests that 4-lane
The 85th percentile
6.3.3 Free Speed Measurements
appeared
(Figure 6.4b).
Moreover,
There was no main effect for sex of the subject
The main effects
values at each level.
slower free speeds than spacious environments
roadside environment
1.2, P>.05, w2=.002).
again in this analysis
and their interaction.
and
type of road variables
at the divided
narrow
(a
6.4 DISCUSSION
The results
ABOVE'
ABOVE.
-
"--
+ 10
0
--
comprehensive
.,..."""-
....•..
--.\
that were tested.
in terms
\
'BELOW
WALLED
ROADSIDE ENVIRONMENT
\
85th
The findings
to provide
from
a
account of the role of these factors in a driver's
of speed.
6.4.1 Presentation
Speed
•............•
BELOW
ABOVE
SPACIOUS
perception
\ ....•...--0
·Mean
1&1
Cl)LIMIT
~
variables
the three sets of data can now be assimilated
SPEED
LIMIT
~
SPEEDr
ZSPEED
+ 10
-10
a.
>
et
:er::!;
j:!:
~
Q
(b)
LIMIT
~
1&1
of the independent
--.85th
.c
E
from this experiment will be discussed
(a)
./
The presentation
speed of the stimulus materials
strongest
effect on both the subjects'
judgements and speed estimates
L
r
D/W
I
DIN
t
I
4L 2L1W
1-----1.
2UN G/W
TYPE OF ROAD
(c)ROAD X ROADSIDE
had the
• Mean
presentation
speeds
were generally
environments
safe operating
speed
in the rural experiment.
Slow
(15 per cent below the posted speed limit)
judged to be too slow for safety in these
and subjects'
Fast presentation
speeds
speed estimates were quite accurate.
(up to 115 km/h), on the other hand,
evoked a too fast response and subjects consistently
under-
estimated these travel speeds.
\"
:/
-10
This result demonstrates
"-\
Mean
.", • Spacious
\
L
,
D/W
,
DIN
,
,
the importance
that sUbjects placed
when making their estimates
safety and travel speed of these road scenes.
\
BELOW
on the speed of presentation
variables
Mean
~Wal,led
4L 2UW 2L/N G/W
TYPE OF ROAD
such as type of road, roadside environment
of the driver were significant
this experiment,
treatment
for considerably
than did presentation
the need for moving stimulus materials
FIGURE 6.4 Free speed variations in km/h for type of road
and roadside environment for rural
roads in experiment 1.
responses
or the sex
main effects or interactions
they accounted
variance
of
While other
speed.
in
less of the
This confirms
when eliciting
perceptual
of safety and travel speed in the laboratory.
This result is not too surprising, given the relative lack
of critical visual features in rural environments.
It will be
interesting
to compare the trends reported above from the rural
experiment with those for semi-rural
the slower posted speeds and the different
environments
may have a marked effect on this result.
- 76 .~
•
.•• <-1'0""
••.
and urban scenes.
Conceivably,
_...-,...,
- 77 -
roadside
f
6.4.2 TYpe of Road
The overall
collector
type of walling normally encountered in rural environments
(roadside trees and forests) does not unduly influence a driver's
particular type of road.
feelings of safety on anyone
finding was that road category
or gravel)
(arterial,
and width of the road surface was critical
for a driver's perception of speed.
Type of road was the second
strongest effect in both analyses, accounting for a sizeable
proportion of the experimental variance.
Speed estimates,
on the other hand were particularly
accurate for spacious environments
This was not influenced
on 2-lane narrow rural roads.
by presentation
a function of the sex of the driver.
The safe operating
this result would influence
speed data showed that as the road
speed, but may have been
It is unclear,
speed perception
however, how
or behaviour
on the
road.
category and number of lanes reduced, subjects' jUdgements tended
towards the less safe end of the scale.
For the error data,
reducing road category and the number and width of the lanes
resulted in less error when estimating travel speed.
In short,
the more major the road, the greater the feeling of safety and
the greater the tendency to under-estimate travel speed.
There was also a suggestion from the free speed data
collected at these sites that a walled environment had some
effect on travel speed for 2-lane highways.
interesting
as it is not really reflected
(there was some sign of a difference
The number of lanes on divided
jUdged less safe than the same type of road with only two lanes.
Moreover, the speed estimates on these wide roads appeared
more accurate
significant).
Moreover,
while there was a significant
in accuracy at estimating
reduction
travel speed for these sites, it is not
clear how this relates to performance
on the road.
This is
another aspect of the rural results that needs further
examination.
than those on 2-lane divided roads, and
free speeds recorded at wide-divided sites were, on average,
krn/h less than at narrOW-divided rural sites.
3
This finding was not expected and, if robust, has important
connotations for the perception of speed.
However, it is
6.4.4 Driver Variables
The subject pool consisted
of equal numbers of experienced
(full driving license) and inexperienced
(first year, provisional
license) drivers as well as males and females.
necessary first to establish the full extent of this rural
anomaly in other environments.
possible
Hence, it was
to evaluate the effects of these two driver variables
the perception
6.4.3 Roadside
data
in the safety estimate
between walled and spacious 2-lane sites but this was not
roads, however, was an
exception to this finding.
For this road category, wide roads
(more than two standard width lanes in each direction) were
slightly
This result is
in the perceptual
in
of speed.
Environment
The amount of driving experience did not have any
This variable had more effect on drivers'
than on their ability to jUdge travel speed.
generally
assessed
fast presentation
between
roadside
much safer than walled
speeds.
safety responses
Spacious sites were
sites , especially
for
There was no sign of any interaction
environment
and type of road suggesting
that the
significant
effect on the results of this experiment.
experienced
and inexperienced
drivers responded
the same way, and this variable has practically
any other factor in the experiment
sign of any experience
and roadside
- 79 -
on
The only
sex, speed
in the speed estimate analysis,
- 78 -
---,.'--------=~.--------
no influence
in both analyses.
effect was in the experience,
interaction
Both
in essentially
but then
with very little power
consideration.
performance
( w2=O)
While driver
variable
for the perception
and not worthy of serious
experience
may be a significant
7.
on the road, it is clearly not so important
THE SEMI-RURAL EXPERIMENT
of speed on rural roads.
A semi-rural environment was defined as a rural setting in a
The sex of the driver, however,
these data.
the safe responses
addition,
showed signs of influence
Sex was close to being a significant
as a between-subject
manipulation.
it did interact with roadside
environment
as well as with type of road and presentation
estimate error data.
on
main effect in
In
in these data
speed in the speed
built-uP area. These environments can be found on the outskirts
of towns and cities where urban design requirement, such as
kerbing and overhead lighting, are included on otherwise rural
roads. In addition, urban roads near some golf courses and those
that travel through undeveloped areas, such as riverines, also
constitute semi-rural environments. These environments are often
subject to speed restrictions, although usually not less than 75
In general terms, female responses
tended towards the less
safe end of the scale than male responses,
tended to under-estimate
Interestingly,
the range of female responses
than males for both the safe operating
estimates
suggesting
less certainty
result was not expected
(Fildes,
in these responses.
from previous
studies
It should be stressed
manipulations
fewer data points than the road and roadside
manipulations.
under certain
effects
While Poulton
circumstances,
in repeated measures
and Freeman
6 SUbjects
designs,
that more subjects are normally
This
in road perception
1979).
that the driver variables
were between-subject
in this
would provide reliable
they nevertheless
always desirable
stated
Moreover,
in these
it would be interesting
to compare
(different sites with the same characteristics) and two
Each site was filmed in a similar manner to that described
would be desirable.
the free speeds of
male and female drivers on the road which, unfortunately,
possible with the experimental design adopted here.
wide, divided-narrow, 4-lane, 2-lane-wide, 2-lane-narrow), two
roadside environments (spaciOUS, walled), two repetitions
presentation speedS (15 per cent above and below the posted
limit). This yielded forty different road scenes in a fully
crossed factorial design experiment.
in each driver variable
in any future experimentation
From the research strategy outlined in Chapter 4, twenty
semi-rural road sites were located within 2 hours of Melbourne
that encompassed the range and levels of the independent
variables of interest. Appendices A-6 to A-B describe the 20
sites selected for the semi-rural experiment, and Figure 7.0
shows some typical examples of these sites.
The independent variables included five road types (divided-
within-subject
(1966) argued that
condition in the rural experiment, these findings must be
considered more tentative than the other results and further
testing
7.1 STIMULUS MATERIALS
with considerably
experiments.
As there were only 9 subjects
A divided semi-rural arterial can include urban freeways
by this definition.
tended to be larger
speed and travel speed
1979; Triggs, Harris and Fildes,
experiment
and this subject group
speed more than their male counterparts.
km/h.
was not
in the previous experiment.
Four experimental films were then
constructed, each containing ten random, 5 sec road sequences
followed by 10 sec of blank film. Presentation order was again
structured to ensure that each film was presented equally in
every serial position.
- 81 - 80 -
","
FIGURE 7.0
SEMI-RURAL SITES
7.2 EXPERIMENTAL
PROCEDURE
An additional
thirty-six
licensed drivers were recruited as
subjects for the semi-rural experiment (nine subjects qualified
in each of the four driver groups, male-experienced,
femaleexperienced,
male-1st year and female-1st
was tested individually
in the laboratory
year).
Each subject
used previously.
7.3 RESULTS
Statistical
omega-squared,
analyses,
involving
analysis
of variance and
were performed on the safe operating
responses and the travel speed estimate
speed data was once more confined to a descriptive
These analyses
Site 35 - Divided, Narrow, Walled road.
again will be discussed
7.3.1 Safe Operatinq
separately
Free
analysis
only.
for uniformity.
Speed Responses
Appendix C-4 lists the statistical
analysis of variance
speed
errors as before.
performed
summary table of the
on these data, while Figures 7.1
and 7.2 show the plots of the results of interest.
There were
two main effects and three interactions that warrant close
inspection
and these are described
effect
ranking).
(w2
Presentation
significant
=.1678).
speed shown in Figure 7.1b was again
for the semi-rural
SUbjects
in order of their strength of
sites
(F(1/32)=82.8,
jUdged slow presentations
p<.OOl,
w2
as too slow,
compared to what they considered to be a safe operating speed and
fast presentations as either safe or slightly too fast. As in
the rural experiment,
the analysis,
the independent
Site 45 - Undivided,
this variable was the strongest
accounting
effect in
for 79 per cent of the variance
variable manipulations.
2-Lane, Narrow, Spacious road.
- 82 - 83 -
due to
.,
85
65~
Z
a:
Q
~
i=
Cl
o
Do
W
U)
,DIN
SPACIOUS
TOO
SLOWSLOW
2LfN
2LfW
4L
.,FAST
, SLOW ENVIRONMENT
FEMALE
WALLED
MALE
(e)
(d)
(c)
85
SLOW
I DIW
,I,ROADSIDE
75
(b)
(a)
65
TOO TOO
FAST
65r
DRIVER
SEX
TOO,PRESENTATION
SPEED
TYPE OF F~~~I
ROAD
751
85r
65
TOO,Fl~~1
TOOl
85
TOO
FAST
(a)
SPEED X ROADSIDE
(b)
TOO
FAST
ROADS X ROADSIDE
t
i
85
85
•••
Z
Walled
w
:lE
w
....•
,oSpacious
Spacious
75
75
4Walled
-- ~.:~ /.:::-=- <---.
Cl
Q
.,
:::J
:;;--
65
65
Q
w
TOO
TOO
W SLOW "
'
FAST
Do
SLOW
U)
PRESENTATION
!
•••
TOO
FAST
w
85
~
a:
SLOW -"
,
D/W
DIN
SPEED
,
4L
,
2L/W
TYPE OF ROAD
(c) Slow
F\~~I
(d) Fast
85
,.Spacious
Do
~Walled
o
W
u.
,
2LfN
SPEED X ROADS X ROADSIDE
Cl
75
751~ --------
:..--
~
U)
65
/0_
.'"
o
TOO~
SLOW I,
..•__
,
D/W
75
65
.:>-<:...:7"-/.spacious
~Walled
'
'
DIN
4/L
2LfW
TYPE OF ROAD
,
TOO~
SLOW "
2LfN
D/W
,
DIN
FIGURE
7.2 Safe operating speed interactions
of interest
semi-rural roads in experiment
2.
I EXPER;ENCED
DRIVING
FIGURE
NO~ICE
EXPERIENCE
7.1 Safe operating
for semi-rural
speed main effects of interest
roads in experiment
2.
-
84
-
I
- 85 -
,
4L
,
2LfW
TYPE OF ROAD
65
TOO
SLOW
•
for
,
2LfN
""'-
The main effect for type of road shown in Figure 7.1a was
also significant (F(4,128)=24.0, P>.001, w2 =.034). While no road
type was assessed as unsafe in this experiment, subjects did
jUdge the speeds on major roads to be more safe than on minor
roads.
This variable
experiment
was the second strongest effect in the
and captured
16 per cent of the treatment variance.
There was a significant
interaction
(F(1,32)=4.5, P<.05), two-lane wide roads (F(1,32)=16.0, P<.001),
and two-lane narrow roads (F(1,32)=7.7, P<.01). This interaction
attracted 4 per cent of the treatment variance.
environment,
between
shown in Figure
presentation
Speed and roadside
7.2a was also significant
(F(1,32)=
10.8, P<.01 w2 = .0015). Simple effects analysis showed that the
effect of roadside environment was significant for slow
presentations
(F(1,32)=20.9,
P<.001),
Errors
Appendix C-5 lists the statistical
summary table of the
error score data while Figures 7.3 to 7.5 show the plots of the
results of interest.
Three main effects and three interactions
are subsequently described in order of their strength of effect.
between type of road and
roadside environment, shown in Figure 7.2b (F(1,32)=8.3, P<.01,
w2=.0076).
Simple effects analysis showed that the only
significant roadside differences were for divided wide roads
The interaction
7.3.2 Speed Estimation
but not for fast
The strongest
effect in this analysis,
main effect of presentation
140.7, p<.001, w2=.1293).
The travel speed of the slow
the variance attributed
to the independent
variables.
Type of road shown in Figure 7.3a was the second strongest
significant effect (F(4,128)=25.4, pC.001, w2 =.0426).
Subjects'
estimates of speed were under-estimated much more for divided and
2-lane-wide
roads than they were for any other road category.
variance
captured 20 per cent of the treatment
in the analysis.
The third strongest effect was the significant
The triple interaction
road and roadside
also significant
between
environment,
presentation
speed, type of
shown in Figures 7.2c and d, was
(F(4,128)=2.9,
P.<.05,
w2
= .0014).
A simple
between type of road and roadside
differences
roads only (F(1,32) = 13.3, 12.6 and 103, P<.001 respectively).
Further analyses for the significant interactions revealed that
any other road type
the effect of the roadside was significant
speeds
However,
(F(1,32) = 9.9, p<.01 and 32.4, p<.001 respectively)
(F(1,32)
=
Speeds
respectively).
This variable combination
cent of the total treatment variance.
(F(1,32)
= 14.9, P<.001).
This interaction was not a particularly strong
effect, accounting for only one-half a per cent of the treatment
variance.
-
BG
-
but not for
1.2, 0.6 and 0.6, p>.05
roadside was only
for 4-lane roads at slow presentation
interaction
shown in Figure
for both divided wide and divided narrow roads
(F(1,32) = 4.3, and 5.4, P<.05, 25.3,
P<.001 and 9.0, P<.01 respectively).
significant
for 2-lane roads for
environment,
7.4c (F(4,128)=8.7, p<.001, w2 =.0128).
Simple effects analysis
revealed that there were significant roadside environment
interaction effects analysis showed a significant speed by
roadside interaction for 4-lane, 2-lane wide and 2-lane narrow
both presentation
(F(1,32)=
presentation scenes was judged accurately, while faster speeds
were under-estimated.
This variable accounted for 61 per cent of
This variable manipulation
presentations (F(1,32)~9.2, P>.05).
This variable combination
captured less than 1 per cent of the treatment variance.
once again, was the
speed shown in Figure 7.3b
- 87 -
accounted
for 6 per
--./
+20
III a:
I0
aIII
~a:
~
U)
W
Q
j::
..4L SLOWER-L---.J.
SLOWER
, FEMALE
.,FAST
(d)
(a)
MALE
(b)
2UW
2UN
SLOW
PRESENTATION
DIN
ZERO DRIVER
SEX
(c)
OF ROAD
+20 SPEED
(e)
SPACIOUS
+20
. WALLED
FASTER.
. TYPE FASTER·
. D/W
FASTER.
FASTER,
-20
ROADSIDE
ENVIRONMENT
ZERO
SLOWER\...
-20
r---~
ZERO
SPEED X ROADS X ROADSIDE
(b) Fast
FASTER
FASTER
(a) Slow
+20
+20
--- --
.....
.c
-
.'
ZERO
E
..
SpaciouS
ZERO~
.::t.
a:
o
a:
a:
-20
1&1
SLOWER
-20
L' DIW
'
DIN
I~
~
SLOWER'-,
'
'
•
4/L
2UW
2UN
TYPE OF ROAD
1&1
.Walled
.•.••..••
r,;" Walled
.~~/.spacious
/'
.'
I
DIW
•
'
'
DIN
4/L
2UW
TYPE OF ROAD
(c)ROADS X ROADSIDE
FASTER
IU)
1&1
Q
+20
1&1
1&1
Q,
U)
ZERO
.---
Walled
-20
SLOWER '-,.\
1
D/W
FIGURE
-20
SLOWER
I
I
EXPERIENCED
DRIVING
FIGURE
I
I
DIN
4L
2L/W
TYPE OF ROAD
I
2UN
7.4 Travel speed errors for the significant 2- and
3- way interactions
for semi-rural roads in
experiments
2.
I
NOVICE
EXPERIENCE
7.3 Travel speed errors of the main effects
for semi-rural roads in experiment
2.
of interest
- 89 -
- 88 -
~_
..
I
2LfN
-~.;.
..
Roadside
main effect
environment
in Figure 7.3c was also a significant
(F(1,32)=14.2,
p<.001,
=.0064).
w2
Spacious road
speeds were slightly under-estimated while the speeds for walled
scenes were substantially under-estimated.
This effect was the
SPEED X ROADS X ROADSIDE(EXPERIENCED)
FASTER
(a) Slow
FASTER
+20
fourth strongest
(b) Fast
+20
The interaction
:c-
S
.lI:
a:
o
a:
a:
~~
Spacious
_.--~~~~Walled
significant
ZERO
',/~
I
•
spacious~~
-20
Z
I
t
DIW
~
er
~
j:
-20[ ~
SLOWER
FASTER
I
I
I
4L
2LIW
DIN
TYPE OF ROAD
I
2L1N
SLOWER
I
•...
I
DIW
/
-..;/
~/'
Walled
I
I
I
DIN
4L 2LIW
TYPE OF ROAD
FASTERI
analysis
shown in Figure 7.4a and b was also
(F(4,12B)=7.0,
revealed
speed, type of road and
pC.001, w2=
.005B).
Simple effects
that the source of this effect was a
significant roadside environment difference for divided narrow
roads at slow presentation speeds (F(1,32)=20.5, pc.001) and for
divided wide, divided narrow and two-lane narrow roads at fast
I
2UN
presentation
speeds
pC.01 respectively).
SPEED X ROADS X ROADSIDE(NOYICE)
(c) Slow
for 3 per cent of the factor
between presentation
roadside environment,
.•~
ZERO
ILl
o
effect, accounting
variance.
(F(1,32)=22.5
with
per cent of the treatment
and 15.4, pC.001,
and 4.9,
w2=.005B, this interaction
attracted
3
variance.
(d)Fast
In
ILl
Q
The four-way
+20
+20
presentation
ILl
ILl
Q,
In
••...~.•.•.•.•.-Walled
ZERO
..~
-20
SLOWER I DIW
-20
DIN
4'L
2L/W
2L/N
SLOWER
TYPE OF ROAD
I
I
DIW
/' ...- ...-
I
DIN
~
••.....•..
I
4L
,~
I
• Walled
/.Spacious
FIGURE 7.5 Travel speed errors for the significant 4· way
interaction for semi-rural roads in experiment 2.
- 90 -
environment,
shown
environment
interaction
was only
significant for novice drivers (F(4,64)=12.0, pC.001) .
Additional simple effects analyses pin-pointed the source of this
interaction to divided wide, divided narrow and two-lane narrow
1_
2L1W 2L1N
TYPE OF ROAD
speed, road and roadside
roads for fast presentation
speeds
and B.6, pC.01 respectively).
was no significant
1_-
between driver experience,
in Figure 7.5 was significant too (F(4,12B)=4.9, pC.01,
w2=.0038).
Further analysis here revealed that the previous
Spacious
ZERO
interaction
speed, type of road and roadside
(F{1,16)=19.5
and 37.6, pC.OOi
By contrast here, however,
road and roadside
interaction
there
for slow
presentation speeds (F(4,46)=1.5, pC.05).
This variable
interaction captured less than 2 per cent of the total treatment
variance.
- 91 -
7.3.3 Free Speed Measurements
Free speed data were collected
in the semi-rural
and standard
experiment.
deviation
of the free speed records obtained at each
site are listed in Appendices
previous
data.
format,
at all 20 selected road sites
The mean speed, 85th percentile,
A-6 to A-8.
no statistical
analysis
In keeping with
was performed
on these
+10
+10
---
Figure 7.6 shows the plots of the mean and 85th percentile
speed deviations
as well as their apparent
The overall mean free speed for all sites was 79.1 km/h or
roads.
speed.
On average,
above the posted
environments.
Walled
overall
was roughly equal to the posted
the 85th percentile
speed was roughly 9 km/h
speed limit for the semi-rural
speed, compared
environmentsl.
road
environments,
shown in Figure 7.6a (the
sites was roughly
to the -4 km/h differential
varied
SPEED\LIMIT
E
zo
~·::~=--"'_=:::=.;;M;;;e;a;;-n-
__
~
a:
for walled
from +13 km/h to +5 km/h above the
BELOW-'
BELOW
SPACIOUS
Type of road shown in Figure
>
In contrast
free speeds less
~
VI
+10
.
SPEED
LIMIT
<- __
."--. "',
••.....•.... ,.':::-...-_.
'"
"'- I I
'-./
ea n
M
Walled
.Mean
Spacious
-10
I
I
-:
I
1
1
I
DIN
41L
2L1W
2L1N
FIGURE 7.6 Free speed variations in kmfh for type of road
and roadside environment for semi-rural
roads in experiment 2.
in
- 93 -
~~-~_~_=
_
I
41L
I
2UW
TYPE OF ROAD
there does not
effect for road category
- 92 ••••
1
DIN
7.Gb appears to have influenced
with the rural findings,
appear to have been any systematic
these semi-rural environments.
__
J
DIW
(clROAD X ROADSIDE
to- ABOVE
BELOW
than the speed limit and narrow road speeds slightly above the
••••
WALLED
ROADSIDE ENVIRONMENT
ca:
~
C
III
III
' •.••85th
'-Mean
TYPE OF ROAD
mean free speed, where wide roads had average
_
'- '.'/ /
-10
DIW
W---;;pIiiii:-iii.-iii."-
/-',
'"
values for
limit.
speed limit.
.
.//
./
SPEED
LIMIT
-10
3 km/h above the
This meant that the 85th percentile
both environments
posted
85th
:E
sites, again, appear to have had slower free speeds
than spacious
mean free speed at spacious
posted
--.
to-
In other words, the average speed across the
20 sites used in this experiment
~
~
0.6 km/h lower than the average posted speed limit for these
semi-rural
----
around the posted speed limit for the roadside
environment and type of road variables,
interaction.
(b)
ABOVE
(a)
ABOVE
I
r..'."",., ..•
r
The apparent
environment,
interaction
between type of road and roadside
shown in Figure 7.6c, SUggests that the effect for
walled environments
was more pronounced
for divided-narrow,
4-
lane, and 2-lane-wide sites.
This result is consistent with the
previous rural finding, except for the divided road result.
Walling effects again were more obvious for 2-lane wide roads,
adding further SUPPort to the notion of an optimum value between
road (or walling) width and free speed.
responses were consistently
laboratory
responses.
the laboratory
different
judged much slower than were the
Nevertheless,
a shift towards more safe in
is also likely on the road itself, albeit of a
(perhaps larger) magnitude.
This result, then, suggests that some of the speed limits
currently applying
of drivers'
in semi-rural
perceptions
zones may be too slow in terms
of what is an appropriate
travel speed.
Indeed, the number of instances where mean free speeds were above
7.4. DISCUSSION
the posted speed limit in this experiment
further demonstrate
that drivers tend to travel in excess of the speed limit in these
These results
factors
of interest
sets
of data.
I
will again be discussed
in this experiment,
situations.
in terms of the
drawing from the three
An increased
speed limit in some of these semi-rural
zones (especially divided and spacious roads) would be justified
to produce a uniform perception
environments.
1,•.
7.4.1 Presentation
I;
Speed
Naturally,
can be demonstrated
disbenefits.
The speed of presentation of the stimulus materials again
exerted mast influence on the subjects' jUdgements in the semirural experiment.
This variable manipulation was the strongest
significant effect observed in both the safety and speed
estimates, capturing between 61 and 79 per cent of the treatment
variance.
This result was almost identical with the previous
rural
findings.
that there are no potential
Once again, this variable
of safe operating
that many of the semi-rural sites included in this experiment
posted speeds of 75 km/h (only sites 31, 35, 43 and 44 had
For these restricted
had
unrestricted
100 km/h limits).
sites, the
presentation
unrestricted
Speeds were 21 to 29 km/h slower overall than their
rural counterparts.
the subjects'
In general
estimates
terms, the
of safety shown by the subjects to these
Speed estimates were less systematic
wide roads produced greater under-estimates
in that
of speed than narrow
roads, but road category had little systematic effect on these
responses.
For both analyses,
type of road was the second
strongest effect observed behind presentation speeds, capturing a
sizable proportion of the total treatment variance.
By contrast with the rural result for this variable, divided
wide roads were jUdged more safe than divided narrow roads (c.f.
Figure's
6.1a and 7.1a).
In fact, the semi-rural
with the hypothesis
roadwidth are critical
result is much
that road category and
factors in a driver's perception
of a safe
It was shown earlier that laboratory responses were Slightly
less sensitive measures of a driver's perception of safety on the
road.
The validation study in Chapter 5 showed that road
operating speed.
This result casts doubt on the previous finding
that divided-narrow rural roads are somehow more safe than
- 94 -
- 95 -
..
:-,',.~~·7~i.--:'_:·~"I:(;;::~\S~~Ir:·;!'1i.!L~:~\~Z'~'.!1:,~.!!i.~~!,!"t;~l'l ..
f·-
road safety
higher the road category and the greater the number of lanes, the
more in accordance
-'~-_.-
influenced
speed and travel speed.
higher the perception
Figure 7.1b shifted further towards the too slow end of the scale
over that observed in Figure 6.1b. This could reflect the fact
rural
7.4.2 Type of Road
moving road scenes.
There was some Suggestion of a general increase in safety
for semi-rural sites, compared with rural sites. The plot in
of safety across comparable
this action should only be taken if it
~~.Ii\l
..~~'ll.
~;~:II.
\P,,~"~l",.~,mt:'_!l~!l'~"',
'"~. !II~~.~"''''~''''~
__ ''''''!~'~'
""'
_
L
"-.
divided wide rural roads, suggesting
that the anomaly reported
for these rural roads may have been site specific.
This warrants
further investigation.
on driving
performance
and width had a direct
was reported
Oppenlander
(1966), Leong
and Hoffman
(1972) and Smith and Appleyard
reported
and, in particular,
contention
environment
The results
with the performance
encompassing
has a special effect on driver's
However, road width, rather than roadside surrounds,
most critical from the data collected here.
Spacious
environments
had their greatest
further evidence
speed.
seem to be
influence
on safe
for 2-lane wide roads, especially
For travel speed estimates,
at
spaciousness was more influential on divided roads, again, at the
faster presentation speeds.
A walled environment appeared to
safety on particular
the immediate
in both data sets has interesting
for speed perception.
influence the perception
support Smith and Appleyard's
that "apparent width"
surrounding
(1968), McLean
(1981).
implications
operating speed judgements
fast presentation speeds.
earlier by
(1966), Vey and Ferrari
here are in general agreement
findings,
!I
I
The finding that road category
influence
f road and roadside environment
environment,
of speed by reducing the degree of
semi-rural
in conjunction
a countermeasure
roads at higher
of the potential
speeds.
This is
for a walled roadside
with other road variables,
against excessive
to act as
speeding on these roads.
7.4.4 Driver Variables
Moreover,
the results from this study further suggest that
the sensory perceptual
behind the proposed
driver's
speed.
system of drivers
relationship
between
may be the mechanism
apparent width and
If this is so, then drivers
of the effect that the road and surrounding
their driving
performance.
scope for unobtrusive
using engineering
This suggests
manipulation
are probably unaware
environment
There was not very much evidence of any driving experience
effect in speed perception.
have on
there is considerable
main effect or interaction
had practically
no strength of statistical
safe operating
For the speed estimates,
countermeasures.
Environment
speed responses
was not a particularly
capturing
in semi-rural
on the
Of interest,
interactions
Roadside
environment,
however,
strong effect in this experiment,
variance.
environment,
in this experiment.
The
speed, type of road and
as well as the interaction
- 96 -
was not a significant
environment.
speed,
For fast presentation
judgements
of experienced
drivers.
While this was a robust effect,
it should not be over
argued that with
emphasised.
Keppel (1982) and others have the chance of a
multiple testing involving many variables,
significant
effect increases
substantially
and one should be
guarded against placing too much importance on these single
occurrences.
Moreover, the consequences of this finding in terms
though, was the effect that roadside
presentation
experience
of
as well as slower free speeds at
had on the other variables
between
in these
speeds and on divided roads only, novice drivers under-estimated
travel speed for walled roadside environments, contrary to the
accurate
environments.
at best only 3 per cent of the experimental
environment
roadside
environment
walled sites did result in greater under-estimates
most of these sites on the road.
association
it did interact with presentation
type of road and roadside
travel speed in the laboratory,
as a
speed data, and
of travel speed on the road
and environmental
There was no main effect for roadside
However,
was not significant
responses.
main effect, although
7.4.3 Roadside
Experience
in the safe operating
between type of
of the experimental
hypotheses
of the problem in interpreting
are difficult
to determine
the speed estimate
data.
Additional
testing may show whether the amount of driver
experience
is important
in the perception
- 97 -
of speed.
because
Unlike the rural
semi-rural
responses.
finding,
the sex of the subject in the
experiment had no statistical effect on the subjects'
Female mean estimates of safe operating Speed and
travel speed were essentially the same as those of males, and the
range of females responses was not noticeably different either.
In short, the earlier sex differences observed in rural
environments were not observed
suggesting either that females
rural environments
for semi-rural environments,
respond differently to males in
only, or that the earlier
particularly robust.
this anomaly.
Further
testing
findings are not
is necessary
to clarify
8.
THE URBAN EXPERIMENT
The third and final experiment
settings.
Urban environments
containing
residential,
consist of built-up
commercial
and, for this experiment,
in this series involved urban
comprised
or industrial
areas
establishments
suburbs of Melbourne.
Spacious and walled urban roadside environments
noticeably
different
to rural and semi-rural
spacious rural setting consisted
developments,
(a minimum
residential
of 3.5m was required
edge of the road to the nearest building
Walled urban settings comprised
or industrial
treed conditions
Urban areas, too, are subject to considerable
While
areas
from the
line) and low fences.
commercial
developments compared to the equivalent
and semi-rural environments.
restriction.
Whereas a
of areas lacking in trees or
urban spacious settings comprised
with wide nature strips
were
settings.
in rural
speed
some sites are posted at 75 and 90 kmlh,
travel'Sl)eeds in the vast majority
of urban areas are limited to
60 km/h. Thus, there were substantial differences in the factors
and levels between urban and rural or semi-rural environments.
8.1 STIMULUS MATERIALS
Using the research strategy outlined in Chapter 4, twenty
urban road sites were located in the Melbourne Metropolitan area
that encompassed
variables
the range and levels of the independent
of interest.
Appendices
selected for the urban experiment,
A-9 to A-11 detail the sites
while Figure
8.0 shows some
typical examples of the urban sites used.
The independent
wide D/W
variables
, divided narrow
2-lane narrow 2-L/N), roadside
two repetitions
and two speeds
limit).
five road types
(divided
environments
(spacious, walled),
(different sites with the same characteristics)
(15 per cent above and below the posted
This produced
in a fully crossed
-
included
4-lane 4-L, 2-lane wide 2-L/W,
DIN,
forty different
factorial
design experiment.
98 - 99 -
speed
road scene presentations
~
FIGURE 8.0
URBAN
SITES
Filming procedures
and sequences
in the urban experiment
were similar to those described
for the rural and semi-rural
experiments.
films were again constructed,
Four experimental
each containing
ten random 5 sec road sequences,
followed
by 10
sec of blank film. Presentation order of the films again was
structured to ensure that each roll of film was presented roughly
equally in each serial position.
8.2 EXPERIMENTAL
PROCEDURE
A further thirty-six
licensed drivers were recruited
subjects for the urban experiment.
subjects allocated
experienced,
As previouslY,
as
there were 9
to each of the four driver groupS, male-
female-experienced,
male-1st year, and female-1st
year. Each subject was tested individually in the laboratory
described earlier using the same procedure as before.
8.3 RESULTS
Undivided,
2-Lane, Wide, Walled road.
Statistical
omega-squared
analysis,
statistics,
involving
analysis of variance
was performed
and
on the safe operating
speed responses and the travel speed estimate errors.
Free speed
data was again not analysed statistically and, therefore, only
described
described
in general terms.
separately.
8.3.1 Safe Operatino
These three sets of results are
Speed Responses
The analysis of variance
data is detailed
in Appendix
statistical
C-6,
summary table of these
while Figure 8.1 shows the
plots of interest here.
There were three main effects and one
interaction that will be described in order of their strength of
statisical
association
(
w2.
ranking) .
Site 54 - Divided, Narrow, Spacious road.
- 100 -
- 101 -
(a)
75
(b)
F1~~1
The main effect of presentation
o~o~o'
85
significant
(F(1,32=115.8,
speed shown in Figure 8.1b
p<.001, w2=.1014).
Slow
presentations were judged to be too slow while fast presentations
were assessed as being too fast. This variable, again, had the
strongest effect in this experiment,
the treatment variance.
65
capturing
56 per cent of all
65
I-
Z
TOO
1&1
SLOW
I D/W
.L
DIN
==
"
1&1
Q
::)
.,
I
4L
I
2LIW
Type of road shown in Figure 8.1a was also a significant
TOO
SLOW
_
2lJN
SLOW
TYPE OF ROAD
TOO
FAST
Q
main effect
'-----...L
FAST
PRESENTATION
(c)
TOO
FAST
operating
SPEED
lane were equivocal
85
1&1
"
75,
.
a:
o
65
p<.001,
TOO
SLOW
L
1&1
LL.
ce
U)
for 25 per cent of
A significant interaction was observed between type of road
and roadside environment, shown in Figure 8.1f (F(4,128=9.3,
65
1&1
D.
effect, accounting
This variable
75
l-
ce
The safe
for 2-lane and divided roads.
had the second strongest
the treatment variance.
D.
U)
Z
p<.001, w2=.0448).
than on 4-lane roads, while both were judged to be faster than on
2-lane roads.
The results for the number and size of the travel
(d)
85
1&1
(F(4,128)=21.1,
speed for divided roads was assessed to be much faster
I
SPACIOUS
ROADSIDE
.J
WALLED
~LE
ENVIRONMENT
DRIVER
TOO
FAST
(e)
/
65
....•..
65
sites
/Walled
/
75
••....
(F(1,32)=2.0,
p>.05).
cent of the treatment
This interaction
variance
accounted
for 9 per
in the urban experiment.
.Spacious
'./
There was a small but significant
environment,
w2
DRIVING
(F(1,32)= 8.2, p<.01
~0019).
main effect for roadside
shown in Figure 8.1c (F(1,32)=4.3,
Spacious environments
p<.05,
were assessed to be safer than
I
-L--...L
EXPERIENCED
showed that for
and 49.8, p<.001 respectively).
There was no significant
difference observed for roadside environment for 2-lane wide
(f)ROADS X ROADSIDE
85
TOO
SLOW
A simple effects analysis
judged to be less safe than spacious sites
TOO
FAST
0_
= .0165).
spacious sites (F(1,32)=8.08, p<.01 and 5.9, p<.05 respectively)
while for 4-lane and 2-lane narrow roads, walled sites were
FEMALE
SEX
85
75/
w2
divided roads, walled sites were assessed to be more safe than
TOO
SLOW
NOVICE
TOO
SLOW
walled environments,
L-l
DIW
EXPERIENCE
DIN
I
4L
I
2L/'U
TYPE OF ROAD
.1
I
although neither were judged to be unsafe.
In terms of its strength of effect, this variable
per cent of the treatment
variance
captured only
and was considerably
smaller
than either of the other two main effects in this experiment.
FIGURE
8.1 Safe operating speed results of interest
urban roads in experiment
3.
for
- 102 - 103 -
.~":
"
.,;
8.3.2 Speed_Estimation
Appendix
C-7 details~the
analysis
summary of the speed estimate
8.3 show the findings
two-way
interactions
statistical
+20
+20
error scores, while Figures 8.2 and
of interest.
Three main effects and four
require detai~ed
in order of their strength
of variance
of effect
attention
ZERO
,0
0
./'"
\
ranking).
(W2
for presentation
was again significant
-20
(F(1,32)=147.0,
and the strongest
p<'001,
SLOWER-.
SLOWER
effect in this analysis
DIw
while fast presentations
variable
captured
tended
to be under-estimated.
although
strongest
experimental
-
only 3 per cent stronger
manipulation.
This
variance
DiN
in this
than the next
In short, the strength of
a:
o
a:
a:
4'L
2L1w
__
2LIN
FASTER
FASTERI
(c)
...•.
,
SLOW
'---FAST
PRESENTATION
TYPE OF ROAD
E
JII:
or slightly over-estimated,
36 per cent of the treatment
analysis,
.c
The speed of slow
.0191).
w2=
was judged correctly
.~
speed shown in Figure 8.2b
....•
presentations
ZERO!
and are described
-20
The main effect
(b)
FASTER
(a)
FASTER
£~rors
SPEED
(d)
+.20
+20
1&1
1&1
this variable
observed
was reduced
in the previous
substantially,
compared to that
•••
Cl
sets of results.
:E
•••
The type of road in Fignre
effect in this analysis
8.2a was also a significant
(F(4,128)=66.3,
main
p<.001, w2 = .0988).
The
en
wide roads was under-estimated,
for 2-lane narrow roads,
speeds were over-estimated.
correctly.
while
Q
1&1
All other
This variable captured
-20
-20
1&1
1&1
travel speed for divided
road speeds were estimated
ZERO
ZERO
D.
en
SLOWER-,
SLOWER-L-I __
....•
,
SPACIOUS
ROADSIDE
33
than in previous
experiments
and was also noticeably
MALE
DRIVER
ENVIRONMENT
(e)
FASTER
per cent of the factor variance
-'- __
WALLED
stronger
and analyses.
+20
There was a significant
roadside
environment,
p<.001, w2=.0775).
more than spacious
interaction
shown in Figure
Walled
between type of road and
8.3c
speed estimates
speed estimates
for divided
123.2 and 46.8, p<.001 respectively),
ZERO
(F(4,128)=80.5,
were under-estimated
roads
(F(1,32)=
but were relatively
estimated for 4-lane and 2-lane roads (F(1,32)=56.5, p<.001; 58,
p<.05 and 17.5, p<.001, respectively).
This variable combination
accounted
for 26 per cent of the treatment
much stronger effect in this analysis
reported
in the earlier
experiments.
- 104 -
.
--~_.-._ _._--..
-20
overSLOWER'-"
EXPERIENCED
DRIVING
variance and is also a
'
NOVICE
EXPERIENCE
than that previously
FIGURE
8.2 Travel speed errors for the main effects of interest
for urban roads in experiment
3.
- 105 -
FEMALE
SEX
The interaction
between presentation
speed and type of road
shown in Figure 8.3a was the next strongest and significant
effect observed
w2 =.004).
in this analysis
(F(4,128)=5.2,
The speed for fast presentations
p<.001,
roads were under-
estimated more than the speeds of slow presentations
FASTER
(a)SPEED X ROADS
FASTER
(b)
SPEED X ROADSIDE
(F(1,32)=106.7,
with w2=.004,
+20
+20
•...
76.9, 82.5, 30.3 and 54.9, p<.001 respectively).
this variable
combination
per cent of the total treatment
captured between
1 and 2
variance.
.c
E
ZERO
•_
___..... •••.••••••••.•.••
Slow
"-
a:
a:
a:
o
w
w
I-
/
SLOWER
•.Fast
~Spacious
Walled
-20
DIW
DIN
4L
2UW
2UN
SLOWER-~
9.8, p<.01, w2=.0016).
(F(4,128)=36.6
SLOW
(c)ROAD X ROADSIDE
FASTER
+20
.·Walled
/
--"'-==-"'-=-"- -.
Female
=--=====Male
responses
p<.05).
l
I
I
I
I
DIW
DIN
4L
2L/W
,
SLOWER
2L/N
This variable
variance
and was not
(F(1,32)=5.2,
affected
revealed that walled
p<.05), whereas
by roadside
female estimates
environment
were not
(F(1,32)= 0.8,
This result, too, was not a particularly
strong effect
value and ranking.
WALLED
ROADSIDE ENVIRONMENT
Sex of the subject shown in Figure 8.2d and the interaction
between driver experience
FIGURE 8.3 Travel speed errors for the significant interactions
observed for urban roads in experiment 3.
too (F(1,32)=4.5,
more than spacious sites for the male
in terms of its omega-squared
L--...l
SPACIOUS
TYPE OF ROAD
for
sex of the subject and roadside
Simple effects analysis
sites were under-estimated
significantly
-20
between
in Figure 8.3d was significant
p<.05, w2=.0014).
••
ZERO
Spacious
-20
sites were more accurate
and 60.3, p<.001 respectively).
The interaction
a.
ZERO
(F(1,32)=
showed that
than walled estimates
attracted only 0.5 per cent of the treatment
an especially strong effect.
environment
SLOWER
Simple effects analysis
(d)SEX X ROADSIDE
+20
U)
speed and roadside
FAST
PRESENTATION SPEED
U)
between presentation
shown in Figure 8.3b was also significant
travel speed estimates at spacious
TYPE OF ROAD
FASTER
environment,
both slow and fast presentations
L-..I _
W
W
W
-
/-,"
•.•..
.-""
:lE
Q
~_
-20
oCI:
t:
••••••__
The interaction
ZERO
.lII:
and sex were not significant
effects
(F(1,32)= 1.5 and 13, p).05, w2 = .0041 and .0026, respectively).
However, they both attracted noteworthy amounts of the treatment
variance
(ie.
previously,
w2 values
with only 9 subjects
relatively
greater
these two variables
than .001).
As mentioned
were both between-subject
in each condition.
factors
Thus, even with
less data points than the within-subject
manipUlations, they still attracted
total treatment variance.
- 106 -
- 107 -
sizable proportions
of the
8.3.3 Free Speed Measurements
Free speed data were again collected
selected urban road sites.
standard deviation
and
of the free speed records obtained at each
site are listed in Appendices
analysis
at each of the 20
The mean speed, 85th percentile
was again performed
A-9 to A-11.
on these data and the results are
once more in terms of apparent
trends.
shows the plots of the mean and 85th percentile
variables,
as well as the interaction
between these two factors.
.c
This was approximately
speed limit
(the arithmetical
limits for all sites).
speed were approximately
4 km/h above the average posted
average
On average,
12 km/h
value of the posted speed
the 85th percentile
free
(roughly 20 per cent) above the
posted speed limit for the range of urban sites included in this
study.
free speeds than comparable
8.4a.
zo
-l-
spacious
The mean free speed value
to yield slower mean
sites, as shown in Figure
LIMIT
LIMIT
itce
>
I-
to the +8 km/h mean speed for
'
WALLED
SPACIOUS
BELOW~I""
ROADSIDE ENVIRONMENT
DfW
DIN
~
+10
III
III
D.
In
/"".-~
,.
/---.-
I
SPEED
LIMIT
I
-10
--\
Mean
• Spacious
\
\
I
\
Mean
·Walled
spacious sites.
The 85th percentile values were +8 and +17 km/h
respectively for these urban road sites.
BELOW
The type of road, shown in Figure
have influenced
free speeds in urban areas.
speeds and 2-lane narrow
speeds were slightly
posted speed limit, while divided
narrow
roads were all faster than the posted
percentile
8.4b, again appeared to
I
1
DfW
I
I
DIN
4L
t
2L1W
I
2UN
TYPE OF ROAD
Divided wide road
less than the
4-lane and 2-lane wide
speed limits.
The 85th
FIGURE 8.4 Free speed variations in km/h for type of road
and roadside environment for urban roads in
experiment 3.
speed was above the speed limit in all cases, varying
from a minimum
of +5 km/h to a maximum
noted that the speed limit on divided
of +18 km/h.
- 108 -
It should be
roads was between 33 and 67
per cent higher (i.e., 100 km/h compared
for other urban road sites.
4L
2LfW
TYPE OF ROAD
(c)ROAD X ROADSIDE
ABOVE
for walled urban sites was close
to the posted speed limit, compared
_Mean
-10
ce BELOW~"
::;
Q
Walled urban road sites again appeared
'.85th
SPEED
Mean
SPEED
,
E
~
The overall mean free speed for the urban sites was 67.9
km/h.
•..•......•..•
85th
values and posted
•...
__
III/ /.--. --.\ \
..•....•..•...
+ 10
Figure 8.4
speed limits for the type of road and roadside environment
(b)
•...•..
•..•.....•....
described
I---~ \\
ABOVE
(a)
ABOVE
No statistical
to 75 or 60 km/h) than
- 109 -
,
2UN
There was some suggestion of an interaction between the type
of road and roadside environment, shown in Figure 8.4c. The
63.8 km/h on average).
effect of a walled
different range of travel speeds between experiments.
As there
were no apparent shifts between the functions reported for the
roadside
environment
appears to be more
pronounced for 4-lane and 2-lane sites than for divided road
sites.
While this result was reported previously in the earlier
rural and semi-rural
experiments,
it is likely to be influenced
here to a large degree by the large differences
speed limits within
in the posted
One would have expected a difference,
therefore, in the absolute
safe operating
level of each function because
speed and travel speed data in all 3 experiments,
it seems likely that there may have been some task influence
work within each experiment.
the urban areas tested.
of the
some slight attenuation
of the laboratory
This is a further suggestion
at
of
in absolute road safety terms as a result
method.
8.4 DISCUSSION
8.4.2 Tvpe of Road
For consistency,
the discussion
of these results will be
structured once again around the independent
from the three sets of urban data collected.
8.4.1 Presentation
variables,
drawing
The type of road, again, exerted a strong influence
safe operating
Speed
the earlier findings,
increasing
environments
in a SystematiC
perception
The presentation
exerted most influence
judgements,
speed.
speed of the moving
on the sUbjects'
and to a lesser degree,
As found in the previous
presentation
two experiments,
speeds tended to be over-estimated.
safe operating
This is further
speed
faster
Slow presentation
speeds, on
to be much slower that the ideal
speed, and their speeds were under-estimated.
support of the importance
One aspect of these results,
experiments
of safety, and a tendency
with
in urban
increase in the subject's
to under-estimate
travel
speed.
In addition, free speeds measured at these sites showed
substantial variations around the speed limit depending on the
Several aspects of this result, however, need to be
road type.
discussed
further.
of movement
for the
however,
found that slow travel
more) were either
was a little puzzling.
that previous
speeds
laboratory
judged correctly
(100 km/h or
or slightly over-estimated
fast and slow presentations
within
each experiment.
higher than in the urban experiment
- 110 -
However,
the
was roughly 65 per cent
(100 km/h compared to only
of these more recent major arterialS.
the particular stimulus materials could have
of the discrepant
urban environments
trend for wide divided roads in
supports the proposition
in road geometry may be influencing
operating
(Hakkinen, 1963; Salvatore, 1968; 1969; Evans, 1970a; Reason,
1974).
Indeed, this relationship was also found here between
mean posted speed in the rural experiment
It was argued that this
induced this result for reasons not readily apparent.
Confirmation
(around 60 km/h) were
while fast travel speeds
roads was reported earlier for
environments.
may have been the result of subtle changes in the design
specifications
Alternatively,
in the introduction
under-estimated,
First, the finding that wide divided roads were perceived
less safe than narrow divided
rural, but not semi-rural,
of speed on the road.
It was reported
generally
safe operating
on their estimates of travel
resulted
the road category
on the
Consistent
speeds were judged too fast for safety and their
the other hand, were perceived
perception
road scenes again
speed and travel speed estimates.
drivers'
that subtle changes
perceptions
of safe
speedS on these roads.
It should also be noted that when filming these roads, the
in the lane that provided roughly an equal
vehicle was positioned
view of the road, left and right of the camera, in the direction
As the camera was aligned with the driver's view (the
of travel.
- 111 -