RSTIBIYL

lllllllllllllllllllllllllllI

3 1293 010379

lljllejlllll

 

 

 

 

 

This is to certify that the

dissertation entitled

"Relations of Driver Understanding of Left— Turn
Display and Driver Age with Left—Turn Accidents"

presented by

Aris Drakopoulos

has been accepted towards fulfillment
of the requirements for

 

 

 

 

 

Ph .D. degree in C1Vll Engineerlng
A \ Mfirrofessor
MSU is an Affirmative Anion/Equal OpporIunity Institution 0-12771

 

LIBRARY
Michigan State
Unlverslty

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity Institution
Wane-v.1

 

 

RELATIONS OF DRIVER UNDERSTANDING OF LEFT—TURN
DISPLAY AND
DRIVER AGE WITH LEFT-TURN ACCIDENTS
BY

Aris Drakopoulos

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

DOCTOR OF PHILOSOPHY

Department of Civil and Environmental Engineering

1993

 

ABSTRACT

RELATIONS OF DRIVER UNDERSTANDING OF LEFT-TURN
DISPLAY AND
DRIVER AGE WITH LEFT-TURN ACCIDENTS

BY

Aris Drakopoulos

This study analyzes older driver comprehension of left—turn signal
displays based on a previously conducted laboratory experiment, and
accident involvement based on field data.

Laboratory data consists of responses of 191 individuals to 82
signal displays representing permitted, protected, red, change and
flashing intervals. Older drivers were found to be more prone to
misinterpretations than other drivers, particularly in response to
complex (defined according to number, color, and type of simultaneously
illuminated signal sections during a particular interval) left—turn
signal indications.

The field database includes information on signal section
arrangement, left—turn phase type, lead/lag phasing, cycle length,
permitted and protected phase duration, number of simultaneously
illuminated left—turn signal lenses, major/minor approach designation,
number, horizontal and vertical signal head position, intersection type,
accident type, driver age and violation, weather, and contributing
circumstances. It consists of 3004 signal, 1217 geometric and 187,715
vehicle accident records.

Laboratory results are used to formulate hypotheses about which
left-turn display characteristics are most likely to affect older driver
safety performance in the field. Left—turn signal displays found to be
less well comprehended by older drivers in the laboratory were generally

found to be present at locations with high older driver left—turn

accident overinvolvement.

The induced exposure method, i.e., the ratio of the proportion of
driver 1 (driver at fault) to the proportion of driver 2 (driver not at
fault--innocent victim) involved in multiple vehicle accidents by age is
used to provide relative accident involvement rates controlled for
environmental and other factors.

Older drivers were found to be over-involved in left-turn
accidents, especially during adverse weather conditions and at
approaches with protected/permitted left-turn control. Left-turn signal
displays were found to be better comprehended and/or be associated with
fewer older driver accidents if: flashing yellow or red ball permitted
indications were used instead of green ball indications; exclusive left-
turn lanes were provided; left-turn signals were placed along the
extension of the left-turn lane centerline instead of between the left-
turn and through lanes, and, protected stacked—three instead of stacked-

four displays were present.

 

Copyright by
Aris Drakopoulos

1993

To my parents,
my wife, and George
who taught me how
not to sleep

 

ACKNOWLEDGMENTS

I would like to express my great intellectual debt to my academic
advisor Dr. Richard W. Lyles for his guidance and patience during the
course of this research. I am also very grateful to Dr. William C.
Taylor for providing advice and support through my years in Michigan
State University. Many thanks are extended to Dr. Connie Page, Dr.
Thomas Maleck, and Dr. Francis McKelvey for their invaluable guidance in
the preparation of this document.

My gratitude to my wife and parents for their encouragement
throughout the years, without which this research could have not been

completed, cannot be overstated.

vi

 

TABLE OF CONTENTS

LIST OF TABLES.... ..... . ................ .... ..... ... ..... ............ix
LIST OF FIGURES. ..................................................... xi

1. INTRODUCTION

HIJFJH
O O O
UIbLuN

O 0

DEFINITION OF THE TRANSPORTATION PROBLEMS

FACING OLDER DRIVERS................... ................. l
DEMOGRAPHIC PATTERNS OF OLDER DRIVERS...... ............... 1
TRANSPORTATION PATTERNS OF OLDER PEOPLE.. ................. 2
TRAFFIC SAFETY ISSUES FACING OLDER PEOPLE ................. 3
SCOPE OF PRESENT RESEARCH................ ...... ...........4

2. LITERATURE REVIEW

2.1.
2.2.
2.3.

MN
0

3. ANALYSIS

INTRODUCTION.................................... .......... 6
OLDER DRIVER ACCIDENT CHARACTERISTICS........... ........ ..7
DRIVER PHYSICAL LIMITATIONS RELATED TO ACCIDENTS.........11
2.3.1. PHYSICAL CONDITION RELATED TO AGE.................12

2.3.2. PHYSICAL CONDITION RELATED TO ACCIDENTS. .......... 13
2.3.3. RELATIONSHIP BETWEEN LABORATORY AND

FIELD MEASUREMENTS.................... .......... ..15
2.3.4. PREVIOUS RESEARCH ON DRIVER PHYSICAL

LIMITATIONS RELATED TO ACCIDENTS........ ........ ..16
DESIGN STANDARDS APPLICABILITY TO OLDER DRIVERS ....... ...18
2.4.1. PERCEPTION-REACTION TIME ................. ..... ...18
2.4.2. TRAFFIC SIGN/SIGNAL DESIGN STANDARDS .. ........ ...19

LEFT-TURN ACCIDENTS, LEFT-TURN PHASE TYPE
AND SIGNAL CHARACTERISTICS...............................22
2.5.1. DRIVER COMPREHENSION AND PREFERENCES

OF SIGN/SIGNAL DISPLAYS................ ......... ..23
2.5.2. RELATIONSHIP OF ACCIDENTS TO LEFT-TURN

PHASE AND SIGNAL TYPE... ..... ......... ............ 26
DISCUSSION OF ACCIDENT EXPOSURE MEASURES......... ..... ...29
THE PRESENT RESEARCH: NEED AND SCOPE... .................. 31
METHODOLOGY
INTRODUCTION.................. ........................... 37
THE LABORATORY EXPERIMENT........... ........ .... ....... ..38
LABORATORY DATABASE RECORDS ............ . ................ .45
LABORATORY DATA ANALYSIS .. .................. . ........ ...46
FIELD DATABASE...................... .................. ...62
FIELD DATA ANALYSIS.. ............... . .......... .. ........ 65

vii

4. ANALYSIS OF LEFT-TURN SIGNAL COMPREHENSION IN
THE LABORATORY

INTRODUCTION.............................................72
ANSWER CORRECTNESS RELATIONS TO SUBJECT CHARACTERISTICS..72
ANALYSIS OF ANSWER CORRECTNESS RELATIONS WITH

VARIOUS SIGNAL CHARACTERISTICS...........................85

5. FIELD DATA ANALYSIS

5.5.

6. SUMMARY,

INTRODUCTION........ .............. ......................118
MANEUVER ANALYSIS.......................................118
LEFT-TURN ANALYSIS......................................136
INVOLVEMENT RELATED TO SIGNAL, GEOMETRIC,

AND ACCIDENT ATTRIBUTES.................................144
5.4.1. SIGNAL ATTRIBUTES................................l49
5.4.2. SIGNAL FACE AND PHASE TYPES......................154
5.4.3. GEOMETRY AND LEFT-TURN SIGNAL POSITION...........1S9
5.4.4. ACCIDENT AND DRIVER CHARACTERISTICS .. ........ ..164
CHAPTER SUMMARY............................ ......... ....172

CONCLUSIONS, AND RECOMMENDATIONS

6.1. INTRODUCTION..................... .......... . .......... ..177
6.2. COMPREHENSION ANALYSIS SUMMARY ............ ... .......... .178
6.3. FIELD DATA ANALYSIS SUMMARY.... .......... ...............183
6.4. CONCLUSIONS.0.00.00.00.00...... .......... O ........ 0.00.0186
6.5. RECOMMENDATIONS................. ..... ...................189
APPENDIX A COMPREHENSION ANALYSIS STIMULI...... ................ ....193
APPENDIX B COMPREHENSION ANALYSIS RESULTS .......................... 204
LIST OF REFERENCES. ..... ...... ...................................... 239

viii

 

LIST OF TABLES

U.S. Driver Intersection Accidents, 1985................. ..... ..8
Perception reaction times in seconds...........................19
Percent correct answers in driving knowledge test scores.......23
Subject demographic and driving experience information... ...... 41

Laboratory database variables..................................46

Tested hypotheses for comprehension analysis part--A... ........ 51
Miles per year by age .......... . .......... ............. ........ 52
Miles per year by age and sex................. ....... . ........ .53
Signal variables used in comprehension analysis .............. ..56

Inter-interval serious error rate comparison...................58

Serious error rate statistics for inter—interval analysis ...... 59
Tested hypotheses, inter-interval comprehension analysis. ..... .59
Tested hypotheses, intra-interval comprehension analysis ....... 6O
Geometries file variables................................. ..... 62
Signal file variables .......................................... 63
Accident variables ............................................. 64
Intersection configuration information ......................... 67

Male driver Dl/D2 table for daytime left—turn accidents at

intersections with permitted left-turn phasing.... ............. 70
Analysis of left-turn signal comprehension-summary ............ 115
Normal signal operation accident involvement (percent). ....... 120

Nighttime signal operations accident involvement (percent)....126

Age cohorts used for figures 5.6 through 5.9... ............... 126
Age cohorts used for figures 5.10 through 5.12 ................ 136
Accident involvement ratios .......... . ..... .......... ......... 145

ix

 

 

5.10.

5.11.

5.12.

5.13.

5.14.

5.15.

5.16.

5.17.

5.18.

5.19.

A1.

A2.

A3.

Bl.

Older driver accident involvement for different signal
controller types..............................................150

Older driver accident involvement for different
cycle lengthSOOOOOOO......OOOOOOOOOOOOO0.00.00.00.00.000......151

Older driver accident involvement for different
permitted phase durations.....................................152

Older driver accident involvement for different
protected phase durations............................. ..... ...153

Older driver accident involvement for different
protected phase signal face configurations..... ............... 155

Older driver accident involvement for different
protected phase types.........................................157

Older driver accident involvement for different
nighttime signal configurations................... ............ 157

Older driver accident involvement for different
intersection types......... ....... . ...... ..... ................ 160

Older driver accident involvement presence of
left-turn laneOOOOOOOOOOOOOO ..... ......OOOOOOOOO0.000000000000161

Older driver accident involvement for different
left-turn signal horizontal positions......... ................ 162

Signal profile configurations................ ................. 163

Older driver accident involvement for different
Signal profileSOOOOOO......OOOOOOOOOOO...0.0.....000.0.0 ...... 163

Older driver accident involvement for different
weather conditions... ...... .................. ................. 166

Older driver accident involvement for different
road surface conditions ................. ...... ................ 167

Older driver accident involvement for different
accident types................ ..... ........... ......... . ...... 167

Older driver accident involvement for different

driver violations........................... .............. ....169
Older driver accident involvement for driver sex .............. 170
Summary of recommendations.. .................. . ....... ... ..... 192
Permitted stimuli..... .................. ...... ................ 194
Protected stimuli.. ............................. . ............. 195
Permitted/protected stimuli. ................................. .198
General comprehension analysis ................................ 204

Table
82. Inter-interval comprehension analysis.........................207

83. Intra-interval comrpehension analysis......... ................ 209

xi

 

LIST OF FIGURES

Figure

3.1. Analysis flowchart. .............................. ... ........... 39
3.2. Typical stimulus display ....................................... 40
3.3. Multiple choice answer box for scene number 1..... ............. 42
3.4. Laboratory data analysis flowchart ............................. 48
3.5. Part B laboratory data reanalysis .............................. 56
3.6. Field database ..... ............ ........... . .................... 65
3.7. Field data analysis flowchart .................................. 69
4.1. Miles driven annually by age.. .......... ..... .................. 73
4.2. Miles driven annually by age (male drivers) .................... 74
4.3. Miles driven annually by age (female drivers) ........ . ......... 75
4.4. Answer correctness by confidence in given answer ............... 78
4.5. Answer correctness by miles driven per year...... .............. 79
4.6. Answer correctness by driver age ............ . .................. 80
4.7. Answer correctness by years driving experience ................. 81
4.8. Answer correctness by sex ...... .. .............................. 82
4.9. All drivers all stimuli serious errors......... ................ 87
4.10. All drivers all stimuli serious error rates .................... 88
4.11. Older drivers all stimuli serious error rates .................. 91
4.12. Older drivers all stimuli correct answer rates ................. 92
5.1. Daytime accidents.... ...... . ...... ..... ....................... 119
5.2. Permitted locations ........................................... 121
5.3. Protected locations ........................................... 122
5.4. Permitted/protected locations ................................. 123
5.5 Nighttime operations .......................................... 125

xii

Liam—re

5.6.

5.12.

5.13.

Daytime accident involvement ratios—through accidents.........127
Daytime accident involvement ratios-right accidents...........128
Daytime accident involvement ratios-left accidents............129
Nighttime accident involvement ratios............... ........ ..130

Daytime left-turn accident involvement ratios-permitted
locations ............. ............................ ............ 137

Daytime left—turn accident involvement ratios-protected
locations.... ...... ...................................... ..... 138

Daytime left-turn accident involvement ratios-prot/perm
locations............ ......... ................................l39

Involvement ratios by intersection maneuver and left-turn
approach control .............................................. 141

xiii

CHAPTER 1
INTRODUCTION
1.1. DEFINITION OF THE TRANSPORTATION PROBLEMS FACING OLDER DRIVERS
The number of "older persons" is increasing faster than any other
age group in the United States today. Likewise, the percentage of older
people that drive is increasing dramatically both because of their
continuing need to travel, since more older people are retiring in the
suburbs where their transportation needs cannot be satisfied by public
transportation or walking, and because they are living longer, healthier
lives, and feel capable of driving. Moreover, certain traffic accident
patterns have been associated with older drivers, such as a larger
percentage of injury and fatal accidents at intersections, especially

those involving vehicles turning left. The trend toward more older

 

drivers in the coming decades makes the study of the causes of such
accidents urgent, since accident numbers are expected to increase
proportionately to the older driver population.
1.2. DEMOGRAPHIC PATTERNS OF OLDER DRIVERS

There is no uniform definition of the term “older driver" in the
literature, and the term has been applied to drivers from 55 to those
over 75. Some authors use the terms "young—old" and "old—old" in
recognition of the fact that a large portion of drivers 55 to 70 years
old have no serious physical limitations and go about their everyday
lives the same way their younger counterparts do, while a larger
portion of drivers over 70 years old experience physical limitations
that may interfere with everyday activities among which is driving. The
"graying of America" has been well established demographically.
Moreover, since younger age cohorts are smaller in size, there will be a
higher percentage of older persons in the population. Thus, the fastest
growing age cohort both in absolute and relative terms is that
containing older persons (80, 104, 139).

Since only a small percentage of the older population is

institutionalized and taken care of with the help of special care

1

providers, attention needs to be given to the majority of older persons
that have no serious physical limitations, live healthy and active
lives, and are dependent on their own means to satisfy their
transportation needs (80, 104, 139).

Available data (104) show that most persons choose to retire in or
near the communities where they lived and worked for most of their
lives. Thus, the migration to warmer parts of the country is not
expected to have significant impacts on the numbers or percentages of
older persons in the majority of states. This point notwithstanding,
there will still be some states that experience greater problems than
others because their climate attracts some older persons from other
parts of the country.

1.3. TRANSPORTATION PATTERNS OF OLDER PEOPLE

The vast majority of older persons continue to rely on the
automobile for their transportation needs such as shopping, visiting
doctors, worshiping, and socializing (80). More older people in the
near future will need and want to use their automobile since the
majority of them will live in suburbs and will, in all likelihood,
continue to make inter-suburb trips much as they do today.

The 1983 National Personal Travel Survey (NPTS) data indicate that
the number of daily trips and annual travel miles for ages 5-15 are 2.3
and 5,800 respectively; these numbers peak at 3.1 and 12,700 between
ages 40 and 49 and decline thereafter so that for people older than 65
years of age they are 1.8 and 4,400, respectively. Despite the decline
in miles driven by age, the number of drivers over 65 years of age and
the average miles driven by this group have been increasing steadily
(78, 139).

The next most popular mode of transportation among the elderly is
walking. Given that the length of walking trips diminishes with age,
the percentage of trips contained within the limits of their
neighborhood increases. Provision of public transit for inter-suburb

trips is typically unrealistic in terms of service and cost (139). It

has also been shown that older persons avoid using public transit
because they perceive it as dangerous in terms of exposure to crime and
unsafe in terms of increased chances for injury as well as exposure to
weather-related and contagious diseases.

Thus, provision of safer and adequate transportation for the elderly
will have to be focussed on the provision of a roadway environment
suited to the needs of the older driver.

1.4. TRAFFIC SAFETY ISSUES FACING OLDER PEOPLE

Given the predominance of the private auto and walking modes among
older people and the steeply increasing number of older drivers, traffic
safety professionals need to focus their attention on providing a safe
environment for older drivers. Certain accident patterns, such as
increasing accident rates, an increasing concentration of driver
fatalities at intersections and changes in manner of collision
associated with advancing driver age have been identified in research
efforts across the country. Older driver physical and mental parameters
have commonly been used as explanatory variables of accident trends
leading to suggestions about a variety of safety countermeasures some of
which fall within the purview of the traffic engineer and can be readily
applied in the field in order to alleviate, to some extent at least,
older driver safety problems. Such countermeasures may include, for
example, the use of larger and/or more luminous signs and signals to
alleviate problems with visual acuity, or provision of longer sight
distances to allow for longer decision—making time.

Within the intersection environment, older drivers tend to be
overrepresented in accidents when attempting a turn, especially a left-
turn. From a traffic engineer's point of view, addressing this accident
category is desirable for at least two reasons: i) the high
concentration of traffic conflicts and serious accidents at
intersections provides an ideal opportunity to benefit more drivers with
safety countermeasures applied within the limited confines of the

intersection; and ii) information on the safety ramifications of certain

 

left-turn control attributes (such as leading or lagging protected
phasing, use and duration of all-red intervals, particular signal lens
arrangements, and signal head positioning) is either lacking or based on
younger driver observations. Thus, the engineer is currently left
without well-documented guidance for making choices associated with
left-turn phasing, especially where older driver safety is an important
concern.
1.5. SCOPE OF PRESENT RESEARCH

Given the severity of left-turn accidents (particularly for older
drivers) and the variety of left-turn signal displays currently in use,
it is unfortunate that relatively limited work has been done on
relations between driver comprehension of different left-turn signal
faces, left-turn accidents, and driver age. The present effort

addresses this void in the current literature in an attempt to bring a

 

better understanding of the causes of older driver overinvolvement in
left-turn intersection accidents. In terms of number of variables
examined and extent of the time period covered, this research is unique
and addressed the need of providing definitive guidance for the traffic
engineer in terms of the proper application of a wide spectrum of left-
turn signal configurations for the purpose of better accommodating the
older driver.

The first stage of the research consists of an extensive
literature review of efforts attempting to shed light on the causes of
older driver accident involvement. Subsequently, the results of a
laboratory experiment designed to measure older driver comprehension of
left—turn displays are analyzed based on information gathered during the
literature search. Hypotheses are formulated about which left-turn
display characteristics emerge as most likely to affect older driver
comprehension. The final stage of the research involves testing these
hypotheses with field data. The extensive field data include accident,
signal/sign configuration, and geometric variables that are compatible

to a considerable extent with the laboratory variables. Thus, testing

hypotheses based on laboratory variables can be readily accomplished
using field data. Based on the results of the hypothesis testing,

recommendations are made about left-turn signal attributes that are

found to be beneficial/detrimental to older drivers.

CHAPTER 2
LITERATURE REVIEW
2 . 1 . INTRODUCTION

A considerable volume of research has been dedicated to older
driver safety issues (5, 20-22, 33, 41, 60, 100, 103, 144, 149). One of
the most crucial issues identified so far has been the higher
concentration of older driver serious accidents within the intersection
environment. The nature of such collisions needs to be examined as a
first step in determining their underlying causes. While accident
information can be used to identify any patterns associated with older
drivers, the investigation into the causes must rely on resources such
as information about older driver driving behavior, driver—perceived
difficulties with the driving task, driver preferences in seeking a
safer driving environment, and driver examination scores.

Thus, citations issued to drivers involved in accidents can
provide helpful leads into why drivers get involved in an accident;
driver ranking of the most serious problems faced in the field can be
used to help the engineer alleviate identified problems to the extent
possible; driver preferences can be kept in mind when designing
intersections; and driver examination scores may be used to assess
relationships between driver comprehension of traffic situations and
probability to be involved in an accident. In addition, commonly held
beliefs about the contribution of older driver physical and mental
limitations to accident experience need to be addressed and assessed
based on the most current information available.

A crucial element in correctly assessing the extent of the older
driver safety problem is the use of an accident exposure measure that
will compensate for peculiarities of older driver travel behavior such
as avoidance of peak hour driving, avoidance of bad weather, and

nighttime driving.

 

2.2. OLDER DRIVER ACCIDENT CHARACTERISTICS

Some uncertainties that surfaced through research into older
driver accident involvement remain to be resolved: although the majority
of accidents involving older drivers has been found to follow patterns
identical to those of other drivers, particular trends emerge for older
driver accidents that, despite concerning a minority of accidents,
become a major source of concern because they are associated with a
higher severity; yet, for another minority of accidents, researchers
disagree on whether differences exist between driver ages.

Indeed, the majority of accidents caused by older drivers are on
dry roads (73) (71), during weekdays, in clear weather, on straight and
level roadways, and more likely to involve local residents (71). Most
injury and fatal accidents occur at non—intersection locations (139,
155). Some disagreement exists among researchers about whether
significant differences are present between daytime and nighttime older
driver intersection accident involvement: three studies report a higher
involvement in daytime accidents (71, 86, 134) while a Michigan study
(73) found no differences between daytime and nighttime accidents.

Many researchers however, agree on a variety of findings (based on a
minority of accidents) that, taken together define what may be construed
as an older driver "accident profile" which sets older drivers apart
from other driver age groups. There is wide agreement, for example, on
the finding of increasing involvement in multi-vehicle accidents with
increasing driver age (4, 52, 71, 85, 86, 145). Consistent with these
studies is the finding that older drivers are less likely to be involved
in single-vehicle rollover accidents (134).

Perhaps the most important accident pattern related to older drivers
is an increase in fatal accident rates with increasing driver age. A
Wisconsin DOT publication (153) identified drivers over 75 years of age
as the second highest ranking driver group (after young drivers) in

fatalities with 4.0 fatal accidents per hundred million vehicle miles

traveled. Verification and explanation of this pattern should be given

 

 

 

a high priority given the dramatic increase in the number of older
drivers expected in the next few decades and the need to come up with
appropriate countermeasures.

In the urban fringe environment where most older drivers currently
live and are expected to continue to live in the coming decades, drivers
over age 64 were found to have a higher involvement in intersection than
mid-block accidents (4, 73) while the highest concentrations of older
driver intersection accidents were found to be at high—volume
intersections controlled by traffic signals or stop signs (77). Not
only were older drivers found to have a higher propensity than other
drivers for intersection accidents, a number of studies found a higher
involvement in injury and fatal accidents in intersections. A
disproportionate concentration of older driver injury and fatal
accidents at intersections has also been noted for 1985 nationwide
statistics (table 2.1.). While 36.7% of older driver fatalities and
59.8% of the injuries occur at intersections, other age groups have only
about 17% of their fatalities and 44-48% of their injuries at

intersections (155).

 

Table 2.1. U.S. Driver Intersection
Accidents, 1985

 

% of Age Group

 

 

Age Injury Fatal
15-25 44.0 16.4
26—64 48.2 17.7

64+ 59.8 36.7

 

 

 

Source: (139 Vol. 2, p 201)

 

 

 

The intersection problems of older drivers are also supported by
the findings of large sample studies in Michigan (73) and Virginia (4)
which used three years of state-wide accident data: drivers over age 60
were found to be highly overinvolved in fatal, and slightly overinvolved
in injury, accidents in intersections.

There have been a variety of attempts to explain the

 

 

 

 

disproportionate concentration of older driver fatal and injury
accidents at intersections. Evans, for example, claims that the pattern
could simply be attributed to increased older driver frailty, claiming
that an impact of similar severity will cause a more severe injury to an
older driver than a younger one due to bone frailty and a lower ability
to recover from injuries. This argument is further supported by
findings that the proportion of intersection fatal accidents climbs to
more than 50% for drivers over 80 years-old drivers (52), and that the
probability of surviving a vehicle crash is reduced significantly after
age 70 and is reduced again after age 80 (86). In summary, according to
this line of thinking, the higher concentration of older driver
intersection fatalities may simply be a consequence of increased frailty
with age.

A closer examination of intersection accidents reveals that other
aspects of the older driver "accident profile" may offer an alternative
explanation (at least in part) for the observed serious accident
pattern: changes in the manner of collision as drivers age show that
older drivers are more likely to be involved in accident types that are
associated with higher injury severity regardless of driver age. While
the majority of intersection accidents involve vehicles moving straight,
there is a shift toward turning vehicle involvement with increasing
driver age that is statistically significant when drivers over age 65
are compared to those 50-64 years old in the urban intersection
environment (4). A Nebraska study supported a similar finding, but
identified statistically significant differences in left-turn accident
involvement only between drivers over 75 years of age and mid-aged
drivers. Although not statistically significant, the percent of crashes
caused by a driver turning left was found to be substantially higher for
drivers over age 65 (25%) than for all drivers (about 12%) (150). The
same study found that left-turning driver accident involvement increased
by 35% (from 17.09% to 23.13%) for older drivers. Similar conclusions

are supported by a study by Garber et al.(4).

 

 

10

The observations regarding the shift from intersection accidents
involving vehicles moving straight to those involving turning vehicles
with increasing driver age are similar to findings reported by others.
Garber et a1. (4) and McKelvey et al.(73) document older driver
underinvolvement in rear-end intersection accidents (involving, for the
most part, vehicles moving straight), and a statistically significant
higher involvement in head-on collisions with a left-turning vehicle.
The latter account for 20% of older driver fatalities according to the
former study. Both of these efforts identified a statistically
significant older driver overinvolvement in angle accidents; they were
responsible for 27% of older driver fatal signalized intersection
accidents (73).

It is reasonable to expect that older driver overinvolvement in
turning, head-on, and angle accidents (that are associated with higher
injury severity than rear-end collisions for all driver ages), will be
accompanied by a correlated increase in older driver injury severity,
regardless of any synergistic frailty effect.

In addition to the already discussed aspects of the older driver

 

accident profile that may be related to the observed increase in
accident severity, other accident characteristics that have emerged from
research efforts include the finding that older drivers are involved in
a higher proportion of backing and parking accidents (71, 73) and that
accidents involving a driver turning right increase by approximately 40%
(from 5.35% for mid—aged drivers to 7.47% for older drivers) with
increasing driver age (150). Older drivers were also found to be
overinvolved in sideswipe-same-direction intersection accidents (4, 73).

Analyses of citations issued to drivers involved in accidents reveal
an emerging older driver "citation profile" that enhances the overall
older driver profile. Older drivers tend to be found at fault by the
officer at the scene of an accident more often than other drivers but
are less likely than their younger counterparts to be cited for

speeding, driving while intoxicated, and following too closely when

11

involved in signalized intersection accidents (134). By contrast, older
drivers are disproportionately more likely to be cited for failure to
yield the right of way and illegal turns--particularly in head-on and
right-angle intersection accidents (4, 71, 73, 90, 115, 153). Finally,
illness, fatigue, glare, and inattention are also commonly cited
contributing circumstances in accidents involving older drivers (75).

Since head-on and right-angle collisions are, as previously stated,
responsible for a large portion of serious intersection accidents
involving older drivers, many efforts attempting to explain the older
driver accident profile have naturally focused on older driver
characteristics that may underlie their tendency to be cited for failure
to yield the right of way and undertake illegal turns in such
situations. In this vein, a considerable body of information about the
decline of physical and mental capabilities with age has been used in
attempts to explain the unusually high occurrences of certain accident
types, citations, and contributing circumstances found for older driver
involved accidents. A detailed presentation of such endeavors is
undertaken in the next section.
2.3. DRIVER PHYSICAL LIMITATIONS RELATED TO ACCIDENTS

Since the highest proportion of drivers with diminished capabilities
is found in the oldest age cohort, accident analyses attempting to
explain older driver accident patterns have often used driver age as a
surrogate for diminished driver capabilities. A significant number of
accident studies have attempted to statistically link physical condition
parameters that are more likely to deteriorate with age. These include
characteristics such as ocular (92) and auditory fitness, increased
perception-reaction time (i.e., the time that lapses from the first
visibility of an event to the initiation of an appropriate response),
increased decision-making time, a higher tendency towards confusion,
inattention, and forgetfulness. Attempts have been made to relate such
characteristics to accident proneness measured in driving violations,

number of accidents, accidents excluding those caused by nonvisual

12

factors, and daytime and nighttime accidents--unfortunately causative
relationships have not been empirically validated.

Although there is consensus among researchers about the fact that
older drivers run a higher risk of being involved in accidents, there
are some fundamental questions about assumptions underlying efforts to
link physical condition parameters with safety measures of
effectiveness. Some of these were summarized by Lerner (65) and are
discussed in the following paragraphs.

2.3.1. PHYSICAL CONDITION RELATED TO AGE

The assumption that physical condition parameters deteriorate
uniformly among all individuals with age is only partially true, since
many researchers have found that there is higher variability in physical
condition parameters among older than younger subjects (107, 152). For
example, certain ocular parameters as well as reaction times of
physically active older subjects were found to be better than those less
physically active (131). In addition, frequent drivers among older
subjects were found to have better reaction times (109); exercise
therapy was found to improve cognitive performance and retrieval
activity (133); and perceptual therapy was found to improve visual
scanning, spatial perception, visual discrimination, and figure-ground
perception abilities (71). A number of research efforts point to an
overall decline in physical performance parameters, such as:

1. A decline in information processing ability starting in
the mid-408 (14);

2. A decline in dual task performance after age 60 (102);

3. A doubling of the time needed for a task involving
extensive peripheral search (83);

4. Two seconds longer to determine what to do when crossing
an intersection (40);

5. Two seconds longer to decide on left-turn safety (51);

6. An accelerated loss of peripheral vision after age 50
(23); and,

7. A decrease in dynamic acuity (older persons are not able
to accurately track objects moving slowly across their
visual field at angular velocities equal to, or greater
than 10 degrees per second).

However, other researchers recognize that individuals "age" at

different rates (112, 152). Wide differences in physical abilities among

 

13

older individuals prompt researchers to loosely define relationships of
driver age to physical abilities. Willis (152) states that "most
individuals will show some decline in physical abilities by age 80," and
that accumulated skills are offset by physiological and cognitive
changes that accompany aging by age 75 (139). Nonuniform loss of
physical ability with age has been the motivation behind efforts on the
definition of a "functional age" (14, 16, 56), that is, a measure of
overall individual performance based on basic skills performance.

In conclusion, it should be stressed that, although researchers
resort to using definitive "older" driver age limits for practical
purposes, these limits do not represent "turning points" after which
driver abilities can be assumed to decline sharply. Nonetheless, the use
of "artificially" defined age groups is valid in view of the fact that,
despite the wide variability in older driver abilities, there is no
disagreement about a general ability decline with age, on average, and
that persons having diminished capabilities are, in fact, more prevalent
in older age cohorts. Comparisons between driver age groups, especially
if such groups have wide boundaries to account for the large variability
within the older driver population, are still meaningful in examining
age effects on driver ability.

2.3.2. PHYSICAL CONDITION RELATED TO ACCIDENTS

Although lower than average physical condition parameters typically
associated with older drivers, such as increased perception-reaction
times, may have some impact on accident experience, studies have shown
that for the most part such relations are quite weak. Older drivers may
compensate to a large extent for the most common problems associated
with aging such as reduced night vision and increased perception-
reaction times by avoiding situations that are more demanding on those
physical parameters where they feel deficient (24, 99). For example,
they often avoid driving at night (in order to compensate for nighttime
vision deficiencies) and drive slower (in order to compensate for

increased perception-reaction time). As an indication of older drivers

14

successfully compensating for reduced physical abilities, accidents that
would be expected to be related to increased perception—reaction times
such as rear-end or fixed object accidents are not prevalent among older
drivers. Also older drivers have reduced involvement to nighttime
compared to daytime interstate accidents (73), presumably because they
limit their nighttime freeway travel. Thus, only older drivers with
better than average physical fitness driver when conditions are more
physically demanding.

The "self-control" mechanism described above explains some
counterintuitive findings, such as the lower involvement of older
drivers in rear-end and nighttime accidents. However, it falls short in
explaining why overall accident rates increase with driver age. A
logical explanation for this apparent discrepancy between theoretical
expectations of lower overall accident rates (if older drivers are
indeed capable of self-control) and the observed increase in accident
rates can be that older drivers may not be aware of limitations in
certain areas of their abilities.

For example, older driver overrepresentation in intersection head-on
collisions while turning left (responsible for a large portion of severe
driver injuries) has been attributed to diminished ability to judge time
to collision and oncoming vehicle speed, and diminished depth
perception. It can be theorized that drivers may be less aware of
decreasing abilities in these areas, while they are more aware of their
nighttime visibility and perception-reaction time limitations.

Regardless of the causes for older driver underinvolvement in certain
accident categories, benefit/cost oriented research should focus on
analyzing the most serious accidents (associated with higher costs) and
revealing their causes. The theory of older driver self-control (if
proven to be true) further supports a concentrated effort on the
analysis of serious accidents. Moreover, the next generation of older
drivers, given their longer and more intense relationship with the

private auto mode and where they are likely to be living (the urban

15

fringe area), may be far less likely to exercise such self control.
2.3.3. RELATIONSHIP BETWEEN LABORATORY AND FIELD MEASUREMENTS

The assumption that laboratory experiments designed to measure older
driver performance are directly applicable to roadway situations may not
always be true. Laboratory experiments often test a single (or at best a
few) variable(s) at a time, while the driving task involves a constant
flow of decision-making based on a large variety of interacting
variables. Contrary to expectations, while laboratory experiments
designed to measure perceptual and cognitive parameters indicate
increased reaction times for older subjects, field tests show no
significant differences between younger and older drivers (49, 61, 87).
The exact compensatory mechanism used in the field is not clear (66).

Those that attempt to explain laboratory and field test

discrepancies point out older driver compensatory mechanisms such as
anticipatory behavior, preprocessing information, driving slower, and
the possibility of limiting the number of monitored information sources
by, for example, allowing for more fixations on the vehicle ahead, not
checking gauges as often. In addition, problems with the definition and
measurement of a "stimulus" and a "response" as well as the large
variability in response time complicate field measurements and might
result in masking the real differences between young and old drivers in
field observations (64). Incompatibility of laboratory test results and
actual field measurements may also be due to measurements of laboratory
variables that cannot be directly translated into measurements of field
variables. For example, visual search time needed to identify particular
patterns among a variety of shapes in a laboratory environment cannot
necessarily be directly translated into visual search time needed in the
field in order to identify a traffic signal, since there are a number of
visual cues that a driver can use to identify the position of a signal
in the field, such as the position of signals in previous intersections
and the behavior of other drivers.

The counterintuitive results mentioned above may also be partially

 

16

explained by the process of "self selection" of subjects involved in
road tests. Perhaps those among older drivers that feel most fit for the
driving task are more likely to participate in road tests compared to
more infirm older individuals, while younger subjects may not be as
hesitant to participate. Such a bias would naturally result in the
diminished differences observed in the field among age groups.

2.3.4. PREVIOUS RESEARCH ON DRIVER PHYSICAL LIMITATIONS
RELATED TO ACCIDENTS

In view of the shortcomings regarding the links between laboratory
tests of driver physical fitness and driver safety measures such as
accident experience and number of violations, it is easy to understand
why a number of studies have failed to establish any statistically
significant relations (20, 21, 22, 37, 84, 135) even in large sample
studies (45, 48, 120). However, certain physical parameters such as low
light recognition threshold, glare recovery, and horizontal extent of
visual field tend to be marginally related to accident experience. Some
positive relations have been identified between dynamic visual acuity
and visual field with accident involvement and driver rate of
convictions (28, 44, 47, 55, 119).

Stronger yet relationships have been shown to exist between
particular accident types, mental demands placed on the driver, and/or
exacerbated driver physical limitations. Studies supporting such
findings are presented in the following paragraphs.

Older driver propensity for accident involvement was found to
increase with increasing complexity of the driving task, for example,
where they have to adjust to other drivers' behavior as is the case in
merging, emerging from minor streets, and turning left at intersections
(74, 77, 126). In addition, multi-vehicle accidents on freeways (where
less interaction with other drivers is necessary) present less of a risk
to older drivers than accidents on non-limited access highways. By
contrast, younger drivers are overinvolved in freeway accidents and

underinvolved in non-limited access roadway accidents (76). Freeway

17

interchanges are a higher risk to older drivers than non-interchange
freeway accidents, presumably because driving on freeway non-interchange
sections is less demanding on quick decision-making abilities while
interchange driving involves interactions with other drivers and is more
demanding on the mental abilities of the driver.

Extent of useful field of view (i.e., the visual field extent needed
for a specific visual task) and mental status test scores were found to
have a statistically significant relationships to accidents in a
57-subject study with an average age of 70 years. The subject’s useful
field of view was found to be reduced when a secondary central task is
added (12, 62), when a target is embedded in distractors, when
similarity between target and distractor is increased, and, finally,
when stimulus duration is decreased. The impact of these factors was
found to be much greater for older adults than for other driver ages
(13, 114, 115).

A model incorporating extent of useful field of view and mental
status score explained up to 40% of the variance in total and
intersection accidents and driver multiple accident involvement for
experiment subjects. Older drivers with restricted useful field of view
and poor mental status test scores had three times more accidents than
those without these problems. Subjects with restricted useful field of
view had 15 times more intersection accidents than those with normal
useful field of view (90, 115). These findings were theorized to be due
to any combination of the following: a reduced speed of visual
information processing, inability to ignore distractors, and, finally,
inability to divide attention (13).

Older drivers with unusually highly restricted physical capabilities
are more likely to be involved in accidents and/or commit more moving
violations. For example, individuals with less than 20/50 binocular
visual acuity and below normal specific contrast sensitivity were found
to have a higher total and nighttime accident rate per million miles

driven (54, 55). Furthermore, a study on 117 impaired drivers 55-88

18

years of age found statistically significant relations between accident
involvement, on-road driving test scores, depth perception, peripheral
vision, reaction time, figure-ground perception, and visual
discrimination (71). In a 1983 study of ten thousand drivers, those with
severe visual field loss in both eyes had accident and traffic violation
conviction rates (an indication of accident involvement potential) twice
those of the general population (55).

In conclusion, increased older driver accident involvement is more
readily explainable for drivers with severely restricted physical
abilities while minor physical ability limitations do not seem to be
linked to accident experience, at least to a detectable degree. In
addition, older drivers are disproportionately represented in accident
types associated with driving situations placing a heavier demand on
driver mental functions such as merging, emerging from minor streets,
and turning left at intersections.

2.4. DESIGN STANDARDS APPLICABILITY TO OLDER DRIVERS

In view of the link between accident proneness of the older driver
and his/her diminished physical capabilities, it has been pointed out
that some standards currently used in highway design, such as
assumptions for gap acceptance, reaction times, and signal/sign
legibility and complexity may adversely affect older drivers (124, 140).

Reviews of the technical basis for a wide variety of standards and
recommendations in the Manual of Unifbrm Traffic control Devices (MUTCD)
identified many standards that had no empirical basis or were based on
empirical data from younger drivers (66, 117). Concerns about the
applicability of highway design and operational standards to older
drivers were also expressed in a study by McGee et a1. (72).

2.4.1. PERCEPTION-REACTION TIME

Existing design standards for perception-reaction times required for
stopping in response to a traffic signal and turning left through
opposing traffic were calculated (71) (table 2.2.) based both on

sequential (upper value) and parallel (lower value) information

19

processing models using data from the literature on older drivers.

 

Table 2.2. Perception reaction times in seconds

 

 

Maneuver Design value Older Drivers
Stop in response to traffic sign 1.0 (106) 1.0 - 2.1
Turn through opposing traffic 2.8 (101) 1.9 3.8

 

 

 

 

 

The somewhat slower older driver perception-reaction times
reported in table 2.2. do not necessarily mean that those drivers are
not adequately provided for by current design values, since various
assumptions built in the current standards allow for some redundancy.
For example, braking distance assumptions incorporate conservative
friction coefficient (travelling at the design speed-wet pavement) and

deceleration rate values (less than that necessary to lock the wheels)

 

that allow for some safety margin. Thus, even if older drivers take a
longer time to react, they may still be provided with adequate distance
to stop safely (64) considering that they are more likely to be driving
slower than the design speed and avoiding adverse weather conditions.
However, the longer reaction times required by older drivers while
turning left through opposing traffic (assuming that the sequential
information processing model is applicable) may place them in serious
jeopardy, especially when the synergistic effect of depth perception
deficiency is taken into account: older drivers turning left may not
only incorrectly assess the distance to and speed of an oncoming through
vehicle, they may also be slow to decide on the appropriate maneuver
required to safely complete the turn or avoid an imminent collision.
2.4.2. TRAFFIC SIGN/SIGNAL DESIGN STANDARDS

Traffic signs are often used to supplement left-turn signal displays
by providing information not immediately evident to the driver by mere
signal placement and appearance, such as the meaning of certain signal
intervals, the traffic movement a signal head is addressed to, or lane
assignment. Supplemental left-turn signs, to be effective, need to: i)

be easily identifiable; ii) be placed in a manner that allows enough

 

20

time for driver response; iii) convey a concise message; and, finally,
v) not overburden the driver with information. It is in this context
that signal legibility distance is examined: longer legibility distances
(i.e., earlier sign detection) allow drivers to perform the necessary
maneuvers in a timely and safe manner. By contrast, short legibility
distances deprive drivers of valuable decision-making and reaction time.
Older drivers will most likely be the first to be adversely affected in
situations allowing brief decision-reaction time windows, since, on
average, they require longer than other drivers to perceive and react to
stimuli.

Since stimulus complexity has been shown to adversely affect older
driver reaction times, the following questions require attention: i) do
the explanatory benefits of a left-turn sign outweigh the additional
information processing burden placed on the driver by the mere presence
of the sign? and, ii) Does the answer to the previous question depend on
sign message/placement and left-turn strategy interactions? Conclusions
of a number of recent research efforts addressing sign relations to
accidents and/or driver age are presented in the discussion that
follows.

Current standards for highway sign legibility assume 50 feet
legibility distance per inch symbol height, which corresponds to a
vision of 20/25 (1). According to Bailey et al.(10) this standard cannot
accommodate 40% of persons 65-74 years of age who have vision worse than
20/25 and 95% of people older than 75 who have 20/40 or worse vision.

Age-related legibility problems are further compounded at nighttime
when legibility distance was found to be 15% less than daytime (7, 38)
and average older driver corrected vision was found to be 20/42 (123).
According to Kline et al.(59), older driver nighttime vision problems
are exacerbated by a higher sensitivity to glare and decreased contrast
sensitivity. Supporting these findings are the conclusions of a
laboratory experiment testing driver response times to signs with

respect to fixed illuminating source type, headlight type, and

 

21

daytime/nighttime illumination where no differences between sexes or
ages were found, but statistically significant differences were
identified in driver response times between day and night; color
specification; and fixed light source type (113).

In a study of sign legibility distance involving 40 subjects with
visual acuity 20/40 or better, younger males had the longest distance
legibility (statistically significantly different than any other age and
sex group). Older and female driver distance legibility was found to be
70%-87% and 84%-92% that of young male drivers respectively. Both
differences were statistically significant (57).

Based on the above mentioned findings about reduced older driver
visibility distance and diminished nighttime visual capabilities, a
number of federally funded highway improvement projects were undertaken
with the explicit goal to improve traffic control device (TCD)
visibility in an effort to provide a safer environment for older
drivers. Typical corrective measures incorporated in these projects
included improved sign and pavement marking reflectivity and the
installation of larger traffic signal lenses and backplates (71).
However, a report summarizing the safety effects of eight such projects
undertaken in the states of Florida, Arizona, and Nevada noted that they
tended to be more beneficial (in terms of accident reduction) to other
ages rather than older drivers in particular (85). Older driver
nighttime accident involvement was lower than that of other driver ages
before the TCD improvement projects, a situation supported by findings
of other accident analyses (73, 76) that showed no older driver
overinvolvement in nighttime accidents. Despite the TCD upgradings, no
significant reduction in such accidents was evident in the after period.

An explanation for this outcome may be that older drivers avoid
nighttime driving as has been documented in the National Personal Travel
Survey (NPTS) (80). Perhaps those older drivers that are at highest risk
to be involved in nighttime accidents are the ones that choose not to

drive at night.

 

 

 

 

22

Presence or absence of supplemental left—turn signs as well as sign
message effects on driver comprehension of particular left-turn signal
displays have been addressed in a number of research efforts. Typical
findings as reported by Hummer (77), are: for protected signal displays,
the presence or absence of a supplemental sign had no significant impact
on driver comprehension; for protected-permissive displays, the message
"left-turn on green or arrow" performed better than the no sign
situation and much better than the "left-turn on green ball" message.

In conclusion, although certain supplemental signs have been found
to enhance driver comprehension of left-turn signals, the presence or
absence of such signs has not been found to contribute significantly to
left-turn accident experience in research efforts to date. Thus,

inability to detect such signs by drivers with reduced visual acuity

 

(the majority of which is found among older drivers) should not be
expected to have a serious impact on accidents related to left-turn
signal/sign displays. The importance of findings of reduced nighttime
vision among older drivers is moderated by the fact that older drivers
seem to avoid nighttime driving and accident analyses show that older
driver involvement ratios are not higher for nighttime accidents.
Avoidance of nighttime driving could be due to older drivers correctly
assessing (even overcompensating for) their vulnerability under such
conditions.

2.5. LEFT-TURN ACCIDENTS, LEFT-TURN PHASE TYPE AND SIGNAL
CHARACTERISTICS

The preponderance of left-turn accidents among older drivers is a
source of concern for traffic engineers seeking to provide a safe
driving environment for all drivers. Traditionally, the choice of
particular left-turn phase strategies has been dependent on traffic
engineering evaluations based on overall accident and delay
considerations ignoring differences in accident experience among age
groups. However, it is widely recognized that existing design standards

may need to be updated and additional criteria may need to be

23

incorporated in the decision making process in order to adequately
accommodate the needs of older drivers.

A number of research efforts discussed in the following sections
have examined driver comprehension and preferences of left-turn signal
displays as well as signal characteristics relations to left-turn
accidents and driver age. Such endeavors may eventually lead to more
comprehensive guidelines about the proper use of left-turn phasing to
better accommodate older driver safety needs.

2.5.1. DRIVER COMPREHENSION AND PREFERENCES OF SIGN/SIGNAL DISPLAYS
Among 2000 subjects over age 55 taking the standard test required to
renew a driver’s license in Nebraska, two predominant patterns were
identified for percent correct answers: first, an overall decline with
age; and, second, percent correct answers for three questions related to
right—angle accidents and four questions related to left-turn accidents

declined with age (table 2.3. source (71)).

 

Table 2.3. Percent correct answers in driving knowledge test
scores (*)

 

Age
55-64 64-74 74+
Right—Angle questions 83% 78% 72%
Left—turn questions 86% 82% 73%

 

(*) Passing score: 80%
Source (71)

 

 

 

Hummer et al. (50) classified questionnaire responses of 402
individuals to eight left—turn signal display understanding questions as
correct, close errors ("actions that would probably not have
catastrophic consequences in traffic"), and gross errors ("actions that
would likely result in a catastrophe in traffic"). It was found that
protected phasing was far better understood than permissive, and
permissive was better understood than protected/permissive. The same
study found no statistically significant comprehension differences in
terms of respondent age, sex, and urban or rural residence. This result

disagrees with McCoy's (71) conclusions in terms of age-related

 

24

comprehension differences. Although McCoy’s sample size is considerably
larger, direct comparisons of the two studies are not appropriate since
they were based on different sets of questions.

Hummer did not find statistically significant relations between
presence of supplemental signs and comprehension of protected displays,
except when a left-turn green arrow was simultaneously illuminated with
a through green ball, in which case the "no sign" situation was found to
be superior to displays using a sign, while for permitted/protected
displays the sign "left-turn on green or arrow" was found to perform
better than the "no sign" condition and much better than the "left—turn
yield on green ball" sign. A simple explanation for the observed
comprehension patterns could be that protected displays are simpler,
relying on a green arrow lens with a unique, concise meaning that needs
no reinforcement from a supplemental sign, while protected/permitted
displays can easily confuse the driver displaying either a left-turn
green arrow and/or a green ball during different intervals. Drivers
might get confused by the message "left-turn yield on green ball" when
the green arrow is simultaneously illuminated with the green ball, not
knowing whether they have the right of way (green arrow message) or they
should yield (sign-green ball message). The information is also
discordant--yield means yield while green ball typically means go.

Hummer’s analysis of driver preferences found that protected is
preferred to permitted and protected/permitted, and protected/ permitted
is preferred to permitted. Preferences were found to be statistically
different by respondent age, with protected/permitted being the top
preference only for 16-25 year—olds. Reasons given by respondents for a
higher overall preference of protected phasing were that it is less
confusing, safer, and results in less delay. Apart from the arguable
accuracy of the perception of lower delay at protected left-turn
locations, drivers' preferences based on ease of comprehension and
safety are in line with the comprehension analysis results based on the

same subjects and accident analyses by other researchers to be presented

25

in the following section.

Finally, Hummer found a preference for leading protected phasing
versus lagging with almost one-fourth of the respondents having no
preference. There were no statistically significant differences in the
preference for leading phasing, except for drivers residing in rural
communities who preferred lagging left-turn phasing.

The most common general problems identified for the intersection
environment in a survey of 425 drivers 75 years of age or older (71)
were: i) right turn on red (RTOR), ii) proper turn lane identification,
and, iii) seeing the signal. Among the same individuals the most common
problems related to left—turns were found to be: i) view blocked by
vehicles in opposite left-turn lane, ii) proper turn lane
identification, iii) identification of acceptable gaps that allow enough
time for a safe turn, iv) identification of left-turn signal meaning,
and v) proper turning path.

The author does not indicate whether the same problems are equally
high in the priority list of other driver ages, however, they can be
directly related to the previously presented studies on older driver
mental and physical abilities.

RTOR presents older drivers with one of the most complex situations
in which the driver has to perform an extensive visual scan of the
intersection to identify any vehicles and/or pedestrians that legally
have the right of way, decide about the adequacy of available gaps, and
finally accelerate and perform the maneuver. The required extensive
physical and mental effort can be reasonably expected to be perceived as
one of the major problems for older drivers at intersections.

Reduced visual acuity and increased sensitivity to glare associated
with aging can be expected to be the reasons behind the difficulty in
identifying proper turn lane and proper turning path, since these
factors are logically related to identifying pertinent traffic control
devices such as lane lines and lane assignment signs.

The laboratory-observed difficulty of older drivers to identify a

26

stimulus in visual clutter or when secondary tasks become more complex
expresses itself in the field as driver difficulty in spotting the
signal among various visual distractions (such as advertising signs) and
mental distractions (such as traffic moving in many directions) present
within the intersection environment.

Reduced depth of field of perception and difficulty in tracking
objects moving across the visual field associated with older drivers can
be readily related to difficulty in identifying acceptable gaps.
Although older drivers are sometimes aware of this situation being a
problem, their overinvolvement in left-turn accidents may be construed
as an indication that they are not aware of the full extent of their
problem. It should be recalled here that, by contrast, in the case of
nighttime driving, older drivers not only seem to be aware of their
limitations, but they seem to be able to accurately assess their risk
and thus often avoid nighttime driving.

2.5.2. RELATIONSHIP OF ACCIDENTS TO LEFT-TURN PHASE AND SIGNAL TYPE

In view of a higher percent of older drivers correctly understanding
the meaning of protected left-turn phasing, and a stated preference for
protected (and particularly leading) left-turn phasing based on
perceptions of improved safety and message clarity, it is interesting to
examine whether better comprehended and more widely preferred left-turn
phase types are actually associated with lower older driver accident
involvement.

A before-after study with approximately one-year before and after
periods (6) based on approaches with left-turn phasing at 24
intersections found that a change from permitted to protected left-turn
phasing in intersections with comparable average annual daily traffic
resulted in a decrease of left-turn accident rates (number of left-turn
accidents per million left-turning vehicles) from 3.76 to 0.86, while
total intersection accident rates were not affected significantly due to
a shift from left-turn (a decline of 85%) to rear-end accidents (an

increase of 33%). Since left-turn accidents tend to be more severe than

27

rear-end accidents, percentage injury accidents declined slightly
(particular information on injury severity was not provided).

A before-after Arizona study based on four years of accident data for
523 intersection approaches with separate left-turn lanes (143) concurs
with the findings of the previous study, concluding that exclusive
phasing is the safest form of left-turn phasing. Conclusions are more
robust since they are based on comparisons of permitted phasing with all
other forms of left-turn phasing and four years of accident data instead
of two. A Michigan study based on four years of accident data (125)
supported the same conclusions for drivers over age 70 while no
differences were found for drivers up to age 70.

The Arizona study found that left-turns across two opposing lanes
with permitted/protected (lagging) phasing had the worst accident rate
(in accidents per million vehicles turning left), followed by
protected/permitted (leading) left-turn phase, and permitted. For left-
turns across three opposing lanes, protected/ permitted (leading) was
the most dangerous, followed by permitted, and permitted/protected
(lagging).

A study that concentrated on accident statistics for older drivers
found that permitted left-turn phasing is present at most older driver
high accident involvement locations (Nebraska) (71). Drivers between 60
and 69 years of age had a higher involvement in head-on left-turn and
rear-end left-turn accidents at permitted left-turn locations (Michigan)
(125).

The Arizona study showed that changes in left-turn phasing on a
particular intersection are directly related to changes in left-turn
accidents. For example, when a permitted approach with two opposing
lanes is changed to protected/permitted, there is a left-turn accident
reduction, while a change from protected/permitted to permitted is
accompanied by an increase in accidents. The only anomaly observed was
that, for approaches with three opposing lanes, a change from permitted

to protected/permitted was accompanied by an increase in accidents.

28

However, many intersection approaches included in the study had short
before or after periods and results cannot be considered conclusive for
all signal phasing changes included in the study.

An accident analysis using four years of accident data from 14
intersection approaches with lagging left-turn phasing and 15 approaches
with leading left-turn phasing found no statistically significant
differences between approaches with leading and lagging left-turn
phasing based on the number of left-turn accidents per million vehicles
turning left. However, left-turn accident rates per million total
approach vehicles were statistically significantly higher for approaches
with a leading phase than approaches with lagging left-turn phasing
(50). A Michigan study (125) based on an extensive accident database
also concluded that lagging left-turn phasing was safer for drivers
between 70 and 74 years of age while no differences were found for
drivers of other ages. Traffic conflict studies have been used to
provide an estimate of accident potential associated with certain signal
phasing sequences in situations where accident data are lacking. A
left-turn conflict analysis based on data from three intersections
showed a dramatic reduction in left-turn conflicts of the order of 69%
to 94% when left-turn phasing was changed from permitted to protected
phasing (6). These findings were expanded in a more recent traffic
conflict study (50) evaluating leading versus lagging left-turn phasing.
Based on observations collected from six intersections in Indianapolis,
the conclusion was that lagging left-turn phasing was associated with
fewer conflicts under similar opposing volumes. Most of the differences
between the two interval sequences were due to drivers turning on the
yellow arrow at the end of the leading phase.

In conclusion, older drivers tend to be less involved in accidents at
intersections with protected left-turn phasing while they are more
involved at intersections with permitted left-turn phasing. Furthermore,
lagging left-turn phasing was found to be associated with fewer traffic

conflicts and lower accident rates per million entering vehicles. Older

29

driver perception about improved safety afforded by protected left-turn
phasing is verified by accident experience observations, however, their
perceptions of leading left-turns being safer have not been found to be
true.

2.6. DISCUSSION OF ACCIDENT EXPOSURE MEASURES

Many traffic accident studies have shown an over-involvement of older
persons in accidents based on various accident rate measures (5, 21, 22,
42, 60, 96, 98, 149). The magnitude and starting point of the "accident
proneness" of older persons has been shown to depend on the denominator
of the chosen accident rate. For example, fatality rates per licensed
driver show an over-involvement for persons over age 65 with men being
involved more than twice as often as women (31, 34, 139) (150—250
fatalities per million licensed male drivers compared to 50-100
fatalities per million licensed female drivers) while the fatality rate
per mile driven shows an identical increase in involvement for both
sexes (34) starting at age 50 with 8 fatalities per billion miles
travelled and reaching 40 fatalities per billion miles travelled by age
85 (1).

The argument of increased frailty of older persons (often attributed
to, for example, reduced bone strength, fracture tolerance and residual
brain dysfunction following initial injury) as drivers and passengers is
widely accepted among researchers (26, 121, 139, 147, 148). It is
theorized that older persons have a much higher probability of being
fatally injured as a result of an accident that would only injure
younger persons.

Thus, the occurrence of accidents where an older driver is fatally
injured is not necessarily a direct indication of impact severity: the
estimated number of crashes of sufficient impact to kill an 80-year old
driver declines dramatically up to age 65 and increases slightly there
after for male drivers, while the decline for female drivers continues
until age 85. Thus, a higher older driver fatality rate is not

necessarily an indication that older drivers are behaving differently

 

 

 

30

that other driver ages, it may simply be a manifestation of increased
older driver frailty. Fatal or injury accident rates should, therefore,
not be used to detect changes in impact severity with driver age without
controlling for increased frailty.

Some biases are inherent in traditional exposure measures, such as

number of accidents per number of licensed drivers or miles drive . The

exposure measure seems to be biased by the fact that fewer older persons
actually make use of their driver's license than younger persons-—making

fewer and shorter trips.

involvement of male versus female drivers perhaps for a similar reason
(more male drivers actually use their licenses than female drivers).

Using exposure measures based on distance travelled, although the
most common measure for accident rates, does not necessarily provide a
fair comparison between age groups in terms of the environment in which
they travel, especially with the older age group in urban areas. The
National Personal Travel Survey (80) shows that older people have
significantly different travel patterns than the rest of the population
(such as avoidance of peak hours and nighttime, trip origins and
destinations in the suburbs, avoidance of bad weather driving).
Moreover, accurate travel estimation differentiated by age groups are
difficult to obtain.

Left-turn accident involvement based on an exposure of left-turning
volume (number of accidents per million vehicles turning left) would be

more appropriate for the present study, but this involvement measure is

may prefer to turn left at intersections with certain types of left-turn
control while avoiding intersections with left-turn control they
perceive as unsafe or they may choose to turn right around the block

more often than other drivers, if they perceive that it is safer to do

50.

 

   

 

31

Involvement statistics based on number of licensed drivers per age
group may similarly be biased because a smaller number of older holders
of drivers’ licenses actually drive.

The intent of the present study is to make use of an accident
involvement measure that will reveal whether older drivers are
significantly under- or over-involved in certain types of accidents
while controlling for environmental exposure per se. Within this
context, the ratio of the proportion of driver 1 (driver at fault) to
the proportion of driver 2 (driver not at fault--an innocent victim)
involved in multiple vehicle accidents by age will be used to provide
relative accident involvement rates controlled for environmental and
other factors.

Driver-at-fault/innocent victim ratios (referred to as D1/D2 in what
follows) provide an ideal older driver accident involvement measure,
since the older "innocent victims'" (02) distribution in the traffic
stream provides, in essence, a relative exposure measure for older
drivers that reflects older driver travel behavior.

Any age group with a ratio of 1.0 is neither over— or under-involved,
since equal proportions of drivers are identified as driver 1 and driver
2. A ratio of less than 1.0 indicates that drivers in this age group are
more likely to be the victim than the cause of an accident (i.e., they
are under-involved in accidents--they are "safer" drivers), while a
ratio of over 1.0 shows that drivers in this age group are more likely
to be the cause of an accident than the "innocent victim" (i.e., they
are over-involved in accidents--they are not less "safe" drivers). Such
an exposure measure can reflect the travel patterns of all age groups,
since it is the ratio and not the absolute numbers of driver 1 and
driver 2 that are of importance.

2.7. THE PRESENT RESEARCH: NEED AND SCOPE

The need to study the source of older driver over-representation in

accidents and take corrective action is urgent since the population of

older drivers is expected to increase even faster than the increase of

32

the older age cohort (139). Moreover, the next generation of older
drivers may well make more and longer trips than the current one. Key
among older driver safety problems is their particular
overrepresentation in the intersection environment, relative to other
drivers.

At the same time, the intersection environment is suitable for
engineering measures designed to correct accident problems because of
the severity and heavy concentration of accidents in a relatively
limited area where corrective measures may benefit a larger number of
drivers. By contrast, mid-block accidents are less serious and dispersed
over a larger area thus, corrective measures for mid-block locations
need to be applied over a larger area to achieve similar accident
savings.

Previous research efforts attempting to explain increasing driver
accident involvement with driver age have had varying degrees of
success. Higher accident involvement by older drivers is most commonly
assumed to be an expression of reduced driver physical capabilities with
increasing age. However, not all physical attributes that were commonly
found to deteriorate with age were found to be significantly related to
accident experience. Older drivers seem to be able to realize the
existence and extent of a number of their limitations (e.g., visual
acuity at night) and successfully compensate for them as mentioned
previously. On the other hand, older drivers do not seem to be able to
recognize the extent of their physical limitations in other areas (e.g.,
detection of motion across their visual field) and are more prone to
accidents associated with such limitations (e.g., left-turn accidents).
While severe physiological limitations have also been found in a number
of research efforts to be associated with older driver accident
overinvolvement, the number of older drivers with such limitations is
small and does not explain the much wider trend of accident
overinvolvement of older drivers. For example, a study by Johnson et

a1. (55) found that drivers with severe visual field loss in both eyes

33

had accident and conviction rates twice those of the general population,
but there were only 196 such drivers among the 10,000 included in the
study.

Laboratory tests measuring the mental capabilities of drivers as well
as interactions of mental and physical capabilities (e.g., extent of
useful field of view) have successfully correlated poor laboratory test
scores with poor performance in the field. The success of such efforts
stems from a close relationship between demands placed on the driver in
real world situations and those tested in the laboratory. Field-measured
older-younger driver reaction time similarities to simple stimuli may be
attributed to younger drivers spending more time monitoring secondary
information in addition to the main task while older drivers concentrate
on the main task. However, as the main task becomes more complicated,
older drivers become handicapped compared to younger ones since task
complexity starts to exceed their processing capacity while younger
drivers can use reserve mental processing capacity previously occupied
in secondary tasks.

In this context, the intersection environment presents one of the
greatest challenges for driver mental capacity due to the presence of
conflicting vehicular and pedestrian traffic movements, and the need for
quick decision-making regarding reactions to signs, signals, other
driver actions and so forth. Within the intersection environment the
through maneuver may be assumed to be the simplest since the driver has
the right of way and only needs to keep safe distances from other cars
and be ready to stop for a changing signal, while turning maneuvers
require much more information and more intense processing of it. The
right-turn maneuver requires the driver to identify both other vehicles
and gaps in other movements having the right of way while simultaneously
watching for pedestrians. The left-turning maneuver is by far the most
demanding on the driver since it requires clear differentiation of the
shape and meaning of a green arrow from a green ball, interpretation of

a protected/permitted sign and signal display, identification of the

34

proper turning path, and, finally, successfully identification of
acceptable gaps in oncoming traffic and the pedestrian stream crossing
the side street.

Identification of the proper turning path is considerably harder
for drivers turning left than those turning right, since, contrary to a
right turn where the driver is guided by the edge of pavement or curb,
during the left-turn operation the driver has guides only at the
beginning and at the end of the maneuver; namely, the centerline of an
undivided crossroad or median edge of a divided crossroad, and the
curved median end. For the central part of a left-turn the driver has no
physical guidance (except where "puppy trails" are provided). Thus,
drivers with limited physical abilities are at a considerable
disadvantage compared to other drivers when turning left.

Judgment of acceptable gaps for a left-turn maneuver is a process
considerably more complex than that facing right-turning drivers, since
pedestrians are further from the turning driver, making judgment of
their speeds harder (especially for persons with limited physical
abilities), and completing the turn requires the identification of
sufficient simultaneous gaps in pedestrian and vehicular traffic. In
addition, an emergency stop to yield for a pedestrian leaves the left-
turning driver exposed to a far greater risk than the right-turning
driver: in the case of a right turn the vehicle may be susceptible to a
rear-end collision, while a left-turning vehicle is not only susceptible
to a rear-end collision, but also to a much more potentially hazardous
right-angle collision with a vehicle moving in the opposing through
direction.

Older driver accident experience is consistent with the above
observations about maneuvers of increasing complexity and the previously
hypothesized older driver accident involvement mechanism. Indeed, there
is a higher concentration of accidents at intersections with advancing
age, and, within the intersection environment, there is a shift to

accidents involving turning vehicles, especially left-turning ones.

35

In this context, the present effort attempts to link driver
comprehension of left-turn signal displays to accident experience and
driver age. It is expected that, since the left-turn maneuver places a
heavy burden on older driver physical and mental capabilities, left-turn
signal displays that are easier comprehended, and thus place less
demands on driver mental capacity, will be related to lower older driver
accident involvement. The results of tests of driver comprehension of
left-turn signal displays in a laboratory are analyzed with respect to
signal display complexity and the message conveyed to the driver. The
data analysis includes direct tests of driver comprehension of left-turn
maneuver right-of—way rules (no physical limitation interference is
allowed through the experiment design). Thus, it is assumed that
comprehension variations among signal displays are a result of mental
capabilities rather than a manifestation of varying physical limitations
among subjects. The design of the laboratory analysis is directed to
identifying signal display characteristics that may place a heavier
mental load on the subject either by conveying a complex message
requiring interpretation, or leading the subject to an erroneous
interpretation of the meaning of a signal display by mere signal
placement and/or color of simultaneously illuminated through signal
sections. Variables examined in the analysis include number and position
of left-turn signal heads, supplemental sign message, signal face
arrangement, number of through signals, and color configurations of
simultaneously illuminated left-turn and through signals.

Once signal characteristics that are related to driver confusion
have been identified in the laboratory experiment, hypotheses are
formulated about signal configurations that are expected to cause driver
confusion and thus be related to higher accident experience in the
field, as well as those that are expected to be better comprehended and
thus be related to lower accident experience. The hypotheses are
subsequently tested against field data collected over a ten-year period

for 324 signalized intersections in Michigan. Database records include

36

accident, signal, and geometric information and allow the identification
of the intersection approach on which a vehicle was moving prior to the
accident. Thus, the specific left-turn signal/sign configuration facing
the driver-at-fault and the innocent victim can be identified and the
hypotheses regarding signal comprehension can be tested.

Considerable compatibility exists between laboratory and field
variables allowing extensive testing of hypotheses. For example, driver
age, signal position, supplemental sign presence and message, left-turn
strategy, flashing (nighttime) operation configuration, existence of
left-turn lane, and signal sections simultaneously illuminated are
included in both the laboratory and field databases. Additional
hypotheses are subsequently formulated and tested based on variables
exclusive to the field database, such as type and duration of change
interval, roadway width and lane configuration, cycle length, left-turn
phase duration, and weather conditions at the time of the accident.

The results of the analyses of laboratory and field data lead to
suggestions about left-turn signal display attributes and strategies
that should be used and those that should be avoided in order to improve

older driver safety in the intersection environment.

 

 

 

CHAPTER 3
ANALYSIS METHODOLOGY

3 . 1 . INTRODUCTION

As has been pointed out in the literature review, although higher
accident involvement by older drivers is most commonly assumed to be an
expression of reduced driver physical capabilities with increasing age,
not all physical attributes (such as reduced visual acuity and nighttime
vision) that were commonly found to deteriorate with age in laboratory
measurements were found to be significantly related to accident
experience. Some older drivers seem to be aware of certain physical
limitations and adequately compensate for them by driving in a
conservative manner (e.g., avoid nighttime and bad weather driving).
Furthermore, no significant differences have been found between older
and other drivers in field tests measuring reactions to simple stimuli
(e.g., braking in response to a leading vehicle braking). However,
laboratory tests measuring driver mental capacity have been more
successful: poor laboratory test scores have been positively correlated
with poor accident experience, perhaps because older drivers are unable
to compensate for intellection and decision deficiencies rather than
physical or reaction time deficiencies. Accident trends reinforce the
notion that older drivers tend to have more accidents than other driver
ages in more complex driving environments, e.g., they tend to have
relatively more serious accidents within the intersection environment
where more decisions need to be made, and especially when executing the
most complex intersection maneuver i.e., turning left. Thus, it is
reasonable to assume that, if indeed driver comprehension is the most
critical link between driver age and accident experience, an ideal
departing point for research efforts attempting to link driver age with
accident experience is left-turn accidents where such relationships will

most likely be the most pronounced.

37

38

Context of current research

Based on these observations of relations between laboratory
measurements and accident experience, the present research focuses on
analyzing older driver comprehension rather than physical limitation
relations to accident experience. The first phase of the present effort
involves an investigation of signal display characteristics affecting
driver comprehension based on laboratory data collected during an
earlier research effort. A second phase involves the analysis of
accident, geometric, and signal field data collected during the present
effort. The latter analysis is focused on the evaluation of the effects
of various signal and geometric characteristics on the accident
performance of older drivers.

Although the data and analyses for the two phases are independent,
an effort is made to use the knowledge gained from laboratory data in
analyzing the field data. Connections between the two databases rely
primarily on a number of variables present both in the laboratory and
field databases. Thus, it is possible to examine, for example, whether
signal characteristics found to be associated with erroneous answers in
the laboratory environment are also present at intersections where older
drivers have higher accident involvement. Compatibility of conclusions
based on the laboratory and field data sets is discussed in the final
chapter. A schematic representation of the relationships between the
two phases of the work is presented in figure 3.1. The present chapter
includes, in turn, the laboratory and field data, hypotheses, and
analysis procedures.

3.2. THE LABORATORY EXPERIMENT

A laboratory experiment to measure driver comprehension of
various left-turn signal head/sign configurations was part of a 1988
study funded by the Federal Highway Administration (FHWA) (122) to
investigate driver comprehension and accident experience in relation to

left-turn signal characteristics. The study, titled "An Evaluation of

 

 

39

 

Figure 3.1. Analysis flowchart.

 

 

 

later-hr!“ M's-sate“
add-this

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[mile-stupor“ mmm Md“
«Indufinwflua \- «unqumnuuan canmumaauubamo
dune-Mellow
Cad-le- MI.
K\\$fl<kumumuuuuuuun
”WWI-Ike
hummus-c
Whit-odd

 

 

 

 

 

Left-Turn Signal Displays" (FHWA contract DTFH61-85-C-00164), was.
contracted to JHK and Associates, which in turn subcontracted Ketron
Inc. to conduct the laboratory experiment.

Driver comprehension was evaluated based on the responses of 191
individuals to a set of slides depicting 82 different signal intervals
representing 17 signal displays, similar to the one depicted in figure
3.2, used in permitted, protected, and permitted/protected left-turn
phasing. (Permitted/protected is used throughout to denote the left-
turn phasing scheme where both a permitted and a protected phase are
present, but pg; in any particular order unless otherwise indicated.)

Subject selection

Subjects were recruited through newspaper advertisements in
Philadelphia, Pennsylvania; Seattle, Washington; Dallas, Texas; and
Lansing, Michigan; and paid $25 to participate in the two-hour
experiment. The research plan was to recruit equal numbers of male and
female subjects for each of four age groups (16-30, 31-45, 46-60, and 61

and older). General characteristics of the subjects in the experiment

are summarized in table 3.1.

 

4O

 

Figure 3.2. Typical stimulus display.

 

 

 

 

 

 

 

 

\

//<s\ \%

 

 

 

 

 

Source: Final report FHWA contract DTFH61-85-C-00164 "Signal
Displays for Left-turn Control"

 

 

 

Experiment protocol

The experiment protocol involved using a set of slide projectors
set up to simulate the color, shape (ball or arrow), and mode of
operation (steady or flashing) of real signal displays (see figure 3.2).
Bach stimulus replicated an interval of one of seventeen left-turn
signal/sign configurations, referred to as signal face arrangements
(SFA) in what follows. Eighty—two stimuli (see appendix A) were shown
to subjects in two prearranged random sequences, not necessarily
following the interval and phase order of individual signal/sign
configurations. Two groups with similar sex and age distributions were

formed at each testing location and each group was assigned one of the

41

 

Fe Me so Fe
FF oF oF FF
NF F F o
m o o o
o M a o
o o F FF
oF FF NF w
o N o o
F a F o
o F o F
o o N n
m FF vF 0F
N o o o
F FF F F
o 0 OF F
o o M «F
o m w o
v F o o
N G F o
o o m o
o o N o
+mM QM-MN «N-MF NF-o

wuzu—awaxw oz_>F¢a m¢<w>

wmm<u oz—mm—t OF use «mum—o ><t mmgm<_¢<> mmoxu< m4<FoF "NFC:

NM cc NM
NF m NF
M F o
N M N
e M F
M F M
m oF FF
N N N
M M N
N e N
o F M
«F 0N NF
0 N F
e e m
o o M
e o M
v N M
o o o
M o F
F F F
o F F

+me me-FF on-w
>44<Dzz< zm>Fao mw4_:

0M em FoF
0F NN 0N
e o m
e o m
M m o
m m N
FF mF MN
e m o
N m m
N m o
M M o
m MN NM
N M q
o N F
F o FF
N N oF
N FN 0N
F q o
F m m
N o o
M o o
xn-o w4<zwu w4<t
xwm

+00
oo-oe
me-FM
OM-oF

+00
oo-ov
me-FM
OM-oF

+oo
oo-oe
me-FM
OM-oF

+oo
oo-oe
m«-FM
OM-oF

mo<

4<FOF

Lau0u-nam

uz_mz<4

.muou-nsm

NDFF<wm

.muou-n3m

w<44<o

.muou.bom

<_:a4mo<4_zd

zo_F<u04

 

 

casumEuONcH wocmflummxm mcw>fluc 6cm owcmmuooemc Dommnsm

.H.m manna

 

 

42

prearranged stimuli sequences at random. This was done to provide some
control over order-of-presentation effects since resource limitations
did not allow for a completely randomized experimental design.

Subjects were provided with a test booklet consisting of a page of
instructions and six pages of response sheets each of which contained a
number of multiple choice answer boxes corresponding one to one to the

presented stimuli (figure 3.3.).

 

Figure 3.3. Multiple choice answer box for scene number 1.

 

 

Scene 1. a) Y N
Number b) Y N
1 c) Y N
d) Y N
e) Y N

2. 1 2 3 4 5

 

 

 

 

 

 

After being given a brief introduction about the purpose of the
experiment, subjects were shown sample slides depicting stimuli (similar
to figure 3.2) and were told to imagine that they were driving in the
left-turn lane and wanted to turn left. They were instructed to answer
yes or no for each slide to each of the following five options for
action:

a) Turn left, you have the right—of—way.

b) Turn left without stopping unless you have to

wait for a large enough gap in the opposing

traffic.

c) Stop. Then turn left when there is a large
enough gap in the opposing traffic.

d) Stop. Then turn left when there is a large
enough gap in the cross street traffic.

e) Stop. Wait until the signal changes to
indicate that you may proceed. ‘

Subjects were to record their answers to item 1 in the box

corresponding to the appropriate scene number on the answer sheet (see

 

 

43

figure 3.3). If, for example, the left-turn stimulus was a left-turn
green arrow, the correct answers to options a) through e) were
respectively Y, N, N, N, N. If the individual's response was N, Y, N,
N, N, then, that response was incorrect. Thus, a subject's answer could
be classified either as "correct" or "incorrect" according to whether
the response agreed with a set of predetermined correct answers.

For each display, experiment participants were asked to provide
the degree of confidence they felt about the chosen answer. The degree
of confidence was coded on a scale from one to five (a score of five
indicating the highest degree of certainty) and corresponds to item 2 in
each response box (see figure 3.3). Subjects were instructed to ask
questions if the color or shape of a lens or word message of a sign were
not clear, but were not allowed to ask questions pertaining to the
meaning of a stimulus. The experimenter would ask if everybody was done
before proceeding to the next slide. Five example slides were used to
familiarize subjects with the method, followed by the series of 82
stimuli of interest to the experiment (only 81 of which are analyzed in
the present effort). Following the stimuli presentation, individuals
were asked to provide additional information on the last answer sheet
about their age, sex, miles driven annually, and years of driving
experience.

Differences between original and current analysis

The original study addressed driver age only in terms of comparing
correct answer rates for each display among age groups. Only displays
showing statistically significant correct answer rate differences among
age groups were discussed (approximately half of the analyzed stimuli
showed such differences). Correct answer rates were found to typically
decline with driver age. Displays were categorized by mode of operation
(normal or flashing) and interval (i.e., permitted, protected, and
protected/permitted operation). Typical conclusions were that older

subjects had: half the correct answer rate as their youngest

44

counterparts for an interval including a protected left-turn display
simultaneously illuminated with a through green ball; one-fourth the
correct answer rate of the youngest drivers for protected/permitted
displays using a fast flashing yellow interval for permitted left turns;
and particular comprehension difficulties with clearance intervals.

No comparisons among alternative signal configurations that can be
used to convey the same message were offered for older drivers, and
there was no attempt to explain comprehension differences between older
and younger drivers. Thus, it was felt that the analysis needed to be
modified if: i)older drivers were to become the focal point of the
analysis, and, ii)signal attributes that are beneficial/detrimental to
older driver signal comprehension were to be identified.

Furthermore, the categorization of driver responses into correct
and incorrect used in the original analysis was deemed inadequate for
the present study since not all signal display comprehension errors are
believed to be equally grave. If, for example, drivers chose to stop
when they had the right of way, the error was considered to be less
serious than if their answer indicated they thought they had the right
of way when in fact they did not. Thus, errors were separated into
minor and serious errors, respectively.

The literature search indicated that driver mental load is an
important factor in explaining differences in the performance of the
driving task among driver age groups. It is therefore reasonable to
expect that signal displays requiring some degree of interpretation are
more likely to be misinterpreted by drivers than those conveying a
simple message. For example, an illuminated green ball section has a
different meaning for the through driver (proceed, you have the right of
way) than a driver turning left (proceed without stopping if there is no
opposing traffic). Thus, some potential exists for turning drivers to
mistakenly assume that they have the right of way, especially at busy
intersections where drivers are distracted by other decisions associated

with the driving task. By contrast, a left-turn green arrow is only

45

used for the left-turn maneuver and the presence of an arrow always
gives the driver the right of way--therefore, confusion of turning
drivers can be expected to be minimal with respect to such displays.
Based on expectations about signal display characteristics that
might contribute to driver confusion, a number of "complexity" variables
were defined. These characteristics include the number, color, and
shape of simultaneously illuminated lenses; signal placement; and number
of left-turn and through signals. All of these may contribute to driver
confusion. A detailed discussion about complexity variables can be found
within the sections discussing displays used for particular left-turn
phases.
In summary, three innovations to the original study are introduced
in the re-analysis of the laboratory data:
i) Emphasis is placed on older driver performance.
ii) Individuals' answers are analyzed based on a

three-level of correctness concept--answers

considered "incorrect" in the previous study are

further categorized into "minor errors" and

"serious errors" depending on whether subjects

incorrectly chose to give away their right of

way or to violate other drivers' right of way
respectively; and,

iii) Stimulus complexity is introduced in the
analysis as an explanatory variable of driver
comprehension.

3.3. LABORATORY DATABASE RECORDS

The database used for the re-analysis of laboratory data consists
of 191 records, one per subject. Data collected in the original
laboratory experiment were recoded to convert each set of five yes/no
answers for each stimulus into a "1" indicating a correct answer, a "2"
indicating a minor error, or a "3" indicating a serious error. Thus,
for each individual, answer correctness level was represented by a set
of 81 digits (ranging from one to three). Each record contains subject
number, age, sex, miles driven per year, years of driving experience,
answer correctness, and a series of 81 digits (ranging from one to five)

representing the degree of confidence each individual felt about each

46

stimulus. The database layout is illustrated in table 3.2.

 

Table 3.2. Laboratory database variables

 

MONTH

DAY

LOCATION

SUBJECT

AGE

SEX

YEARS OF DRIVING EXPERIENCE

MILES DRIVEN PER YEAR

CORRECTNESS OF ANSWER TO QUESTION 1

CORRECTNESS OF ANSWER TO QUESTION 82
CONFIDENCE IN ANSWER TO QUESTION 1

CONFIDENCE IN ANSWER TO QUESTION 82

 

 

 

3.4. LABORATORY DATA ANALYSIS

Two major directions were identified in the literature review for
studies attempting to explain the poor driving performance of older
drivers: a number of authors cite a general deterioration of physical
abilities, while others believe that it is mainly caused by a
deterioration in mental abilities. Laboratory experiments measuring
mental abilities have shown better correlations with poor driver
performance in the field than similar experiments measuring physical
abilities. Thus, the present research is geared towards identifying
signal attributes that tax driver mental rather than physical abilities.

Indeed, the experimental protocol precluded any consideration of
physical limitations (individuals were free to ask questions about the
physical appearance of stimuli--e.g., wording of sign messages, shape of
lenses) while it measured driver mental load (albeit indirectly), by
measuring answer correctness and confidence in given answer. The
analysis of signal characteristics that contribute to better/poorer
driver comprehension of signal displays in the laboratory environment
provides useful insights into which signal characteristics could be
important in improving older driver signal comprehension in the field.

A correct answer, especially one given with a great degree of

47

certainty, is an indication that a signal message is concise and will
not require a great share of driver attention to be understood in the
field, allowing the driver to concentrate on the other tasks involved in
turning left and contributing to diminished risk of an accident. By
contrast, a hard-to-understand or ambiguous display may lead to driver
confusion or incorrect action and will require a larger share of
available driver mental processing capacity--diverting attention from
other crucial tasks involved with turning left, possibly leading to an
accident.

The main interest in the laboratory data analysis is identifying
signal display characteristics that affect older driver comprehension.
However, before the more detailed analysis of such characteristics, an
examination of subject demographic information is used as a starting
point to establish background statistics about the experiment subjects.
Overall relations between answer correctness (regardless of stimulus
meaning and signal display characteristics) and various driver-related
variables are analyzed to establish a base for signal comprehension,
with which the detailed analysis can be compared. Thus, questions such
as whether older drivers show significantly decreased comprehension,
whether more experienced drivers show better comprehension, and whether
any comprehension differences exist between sexes can be addressed
before undertaking a more detailed examination of which signal meanings
or particular signal display characteristics affect driver
comprehension.

The analysis of laboratory data is divided in two major parts: a
general introductory analysis focusing on demographic and general driver
comprehension-related information, performed in part A, and a detailed
analysis of driver comprehension relations with signal meaning,
characteristics, and driver age, performed in part B. This analysis is

graphically presented in figure 3.4.

48

 

Figure 3.4. Laboratory data analysis flowchart

 

 

 

 

General demographics analysis

 

 

Part A Answer correctness related to
confidence in given answer, age,
sex, miles driven per year, driving
experience

>6 I

Answer correctness relations with
Part 8 various signal characteristics

 

 

 

 

 

 

 

 

 

 

 

The issue of designating a "cutoff age" for the purpose of
defining "young" and "old" drivers has been extensively discussed in the
literature review. As has been mentioned, although mental and physical
abilities associated with the driving task generally deteriorate with
age, a large variation in mental and physical abilities exists among
individuals, even within the same age group. Thus, age cutoff limits
are employed for practical purposes only; they are used to study broader
trends related to driver age and are not related to specific physical
and psychological changes that take place at certain precise ages. The-
limited number of subjects available, and the need to have adequate age
group sample sizes for statistical inferences, allowed for consideration
of four age groups: 16-30, 31-45, 46-60, and older than 60 (the same age
groups as used in the original analysis). The age cutoff for the older
group (60 years) is compatible with typical cutoff ages for "older"
drivers used by the majority of authors which range from 55 to 65 years
of age. Thus, the results here can be easily compared to previously
published findings.

Subject characteristics analysis

The first part of laboratory data analysis involves a broad
overview of subject characteristics and their relationship to answer
correctness. Overall trends in miles driven per year by age group are
examined to test whether subject sample characteristics are consistent

with national trends towards less miles driven per year as drivers age

49

reported in the 1983 National Personal Travel Survey (80); sex
differences in miles driven per year are also examined.

Although the number of subjects (here) is very small for
inferences about the entire driving population, any differences in miles
driven by age and sex among study participants have significant
ramifications for the appropriateness of the use of traditional million
vehicle mile-based accident rates, particularly when investigating
differences among driver ages and sexes: if older drivers drive
significantly more (less) miles than other drivers, overall accident
rates based on accidents per million vehicle miles need to be adjusted
downwards (upwards) to correct for driver age.

It is also important to know whether subjects that felt confident
about their answers were more likely to have chosen a correct answer or
not. Ideally, correct answers should be chosen with a high degree of
certainty indicating that the message conveyed by signal displays is
concise, and erroneous answers should be associated with a high degree
of uncertainty, indicating the existence of some confusion about the
correct signal message. A worst case scenario would involve serious
error answers chosen with a high degree of confidence, indicating that a
signal conveys precisely the wrong message--a situation that should
clearly be avoided. Thus, relations between subject confidence in the
given answer and answer correctness will be addressed in this part of
the analysis. Answer correctness is addressed in terms of correct,
minor error, and serious error percentages among all answers for each
individual. The corresponding terms used throughout this text are
correct, minor error, and serious error rate, respectively.

It is reasonable to expect that the more miles individuals drive,
the better their comprehension of signals will be, since they have more
opportunities to familiarize themselves with the driving environment.
The question arises about whether miles driven per year can be used as a
predictor of how well an individual comprehends left-turn signal

displays. If subjects reporting a low number of miles driven per year

 

 

 

 

50

have reduced comprehension scores, drivers in the low mile per year
category could be given special attention, such as additional signal
comprehension instruction.

Finally, the more experienced drivers are, the better their
comprehension of signals can be expected to be. However, older drivers
that are most likely to have the highest number of years of experience
are expected to show low comprehension scores according to the
literature review. Thus, it is of interest to investigate statistical
relationships between driver comprehension and years of driving
experience in order to identify the age at which comprehension peaks due
to increasing years of driving experience without noticeable adverse
comprehension impacts due to advancing driver age.

Research hypotheses and statistical methods-general comprehension analysis

Database information is presented through general descriptive
statistics, and appropriate null and alternative hypotheses (presented
in table 3.3) are tested for statistical significance. Typical testing
procedures are presented below while a full presentation of results can
be found in chapter 4.

As stated above, older subjects are expected to have reported less
miles driven per year than younger ones. General descriptive statistics
will be used to provide an overview of database information before
proceeding to verify whether statistically significant differences exist
between age groups. An example of null and alternative hypotheses to be
tested is illustrated below:

Pg: There are no statistically significant differences in miles
driven per year among driver age groups.

Hg Miles driven per year differs at a statistically significant
level among driver age groups.

These hypotheses are tested using an analysis of variance (ANOVA)
model where miles driven per year is the dependent variable and driver
age is the independent variable. A 95% confidence interval is nominally

used as the criterion for statistical significance throughout this

 

51

 

Table 3.3. Tested hypotheses for comprehension analysis--part A

 

H : There are no statistically significant differences in miles driven per year
among driver age groups.

H,: Miles driven per year differ at a statistically significant level among
driver age groups.

 

H : There are no statistically significant differences in miles driven per year
among driver sexes.

H1: Miles driven per year differ at a statistically significant level among
driver sexes.

 

H : There are no statistically significant differences in miles driven per year
by each age group among sexes.

H,: Miles driven per year by each age group differ at a statistically
significant level among sexes.

 

 

 

H,: Correct answer(serious error) rates are not statistically significantly
different among miles driven per year categories.

H,: Correct answer(serious error) rates are statistically significantly
different among miles driven per year categories.

H,: Correct answer (serious error) rates are not statistically significantly
different among driver experience groups.

H1: Correct answer (serious error) rates are statistically significantly
different among driver experience groups.

H,: Correct answer (serious error) rates are not statistically significantly
different among sexes.

H1: Correct answer (serious error) rates are statistically significantly

 

 

different among sexes.

 

analysis. A high F-statistic value supports a rejection of the null
hypothesis with a high degree of confidence and indicates that, for the
above example, significant differences in miles driven per year among
age groups exist, and H; is rejected. Conversely, a low F-value
indicates that no significant differences exist among age groups and H0
cannot be rejected. The hypotheses to be tested in this manner are
shown in table 3.3. The Statistical Package for the Social Sciences
(SPSS) is used to compile ANOVA tables for each of the hypotheses listed
in table 3.3. An example is shown in table 3.4.

In table 3.4 the F-statistic value of 1.772 for factor age

52

 

 

Table 3.4. Miles per year by age
Sum of Mean Signif
Source of Variation Squares DF Square F of F
Main Effects 942855462.635 3 314285154.21 1.772 .155
AGE 942855462.635 3 314285154.21 1.772 .155
Explained 942855462.635 3 314285154.21 1.772 .155
Residual 26605785789.5 150 1773719OS.26
Total 27548641252.1 153 180056478.77
191 Cases were processed.
37 Cases ( 19.4 PCT) were missing.

 

 

 

corresponds to a significance of 0.155. That is, the statement that

miles driven per year are statistically significantly different among
age groups is correct only 84.5% (loo-15.5) of the time, which is lower
than the desired 95% certainty, thus Ho cannot be rejected.

A similar ANOVA model can be used to test whether vehicle miles
driven per year are statistically significantly different between sexes.

Once basic age and sex differences in miles driven per year have
been assessed, the interaction between age and sex can be investigated
in order to decide whether sex is differentially related to miles driven
per year according to subject age. The hypotheses to be tested are
stated as:

Hfi There are no statistically significant differences in miles
driven per year by each age group among sexes.

Hg Miles driven per year by each age group differ at a
statistically significant level among sexes.

A multivariate ANOVA model with age and sex as the independent
variables and miles driven per year as the dependent variable will be
used to test the hypothesis.

In table 3.5 the F-statistic value of 2.095 obtained for age
corresponds to a significance of 0.103, an insignificant value. The F-
statistic value of 2.095 obtained for the effect of sex corresponds to a
significance level of 0.002 which indicates statistically significant

differences among sexes. The combined effect of age and sex has an

 

 

 

513
Table 3.5. Miles per year by age and sex
Sum of Mean Signif
Source of Variation Squares DF Square F of F
Main Effects 2580676750.55 4 645169187.64 3.918 .005
AGE 1035217075.90 3 345072358.63 2.095 .103
SEX 1637821287.91 1 1637821287.9 9.945 .002
Z-way Interactions 923426094.556 3 307808698.19 1.869 .137
AGE SEX 923426094.556 3 307808698.19 1.869 .137
Explained 3504102845.10 7 500586120.73 3.040 .005
Residual 24044538407.0 146 164688619.23
Total 27548641252.1 153 180056478.77
191 Cases were processed.
37 Cases ( 19.4 PCT) were missing.

 

 

F-statistic of 1.069 and is associated with a significance level of
0.137, indicating that sex effects are not significantly different
across age groups.

Other analyses will be focused on examining relations between
driver comprehension scores and certainty in the given answer, subject
age and sex, miles driven per year, and years of driving experience.
Statistically significant comprehension differences among age groups are
similarly tested.

In order to simplify the analysis, number of miles driven per year
are classified into four categories: up to 7,000 miles, 7,001-10,000,
10,001-15,000, and more than 15,000; and driving experience is
classified into four categories: up to 12, 13-24, 25-38, and over 38
years of experience.

Answer correctness relations with various signal characteristics

The second part of the laboratory data analysis (part B--figure
3.4) is concerned with determining whether comprehension differences
exist among driver age groups based on stimulus meaning. Although it is
anticipated that, in general, older drivers will have lower correct
answer rates and higher serious error rates, it is also expected that

for certain stimuli there may be particularly pronounced comprehension

differences among age groups while comprehension differences may be less

54

pronounced among age groups for other stimuli. Displays that are
particularly difficult for older drivers to comprehend need to be
identified since they pose a threat to their safety, and the confusing
displays should be avoided to the extent feasible. A broader analysis
of driver comprehension differences based on signal interval and driver
age precedes the more detailed investigation of which stimulus, among
alternatives used for the same interval, is best understood by older
drivers. Thus, the first analysis stage involves comprehension
comparisons among stimuli used for the permitted, protected, and
protected/permitted left-turn intervals, the red interval, the change
interval, and flashing mode operation in order to identify which signal
interval(s), if any, is (are) more likely to be misinterpreted by older
drivers.

Since comprehension scores are compared between groups of stimuli
used for different signal intervals, this component of part B
comprehension analysis is called the inter-interval analysis.
Comparisons are based on comprehension scores (i.e., correct, minor
error, and serious error rates) compiled for the groups of stimuli
belonging to each interval for each individual. A correct answer rate,
a minor error rate, and a serious error rate are compiled for each
individual for all stimuli representing a red interval. A separate set
of answer rates is compiled for all stimuli representing permitted
intervals, and so on. The results of this analysis will determine
whether particular emphasis needs to be directed to specific signal
intervals. The segregation of stimuli into stimulus meaning categories
allows the rank ordering of subsequent inquiries: low comprehension of
the red interval, for example, is more likely to lead to serious
accidents since drivers will violate another driver's right of way,
while low comprehension of a protected left-turn arrow may not have such
dire consequences, since drivers facing such displays already have the
right of way.

The last component of part B of the comprehension analysis

 

 

55

involves comprehension comparisons within groups of stimuli used for the
same signal interval in order to identify which display among
alternative signal displays better conveys the same message to drivers,
particularly older ones. Since comparisons are made within groups of
stimuli conveying the same message, this part of the analysis is
designated as the intra-interval analysis. The detailed stimulus
analysis allows an examination of the effect of supplemental signs,
signal head position, signal section arrangement, and other stimuli
characteristics that may benefit/handicap the comprehension of
particular intervals while they may have no effect (or even the opposite
effect) on other intervals.

Based on the literature search, it is expected that increasing
display complexity will be inversely related to older driver
comprehension. Thus, seven traits of signal display complexity are
examined in this analysis:

1) The simultaneous illumination of two sections on the left-turn
signal face (ILLUMINATED-—e.g., a yellow ball simultaneously
illuminated with a green arrow);

2) The concurrence or discordance of the messages of two
simultaneously illuminated sections on the same left-turn signal
face (AGREELT);

3) The concurrence or discordance of the illuminated left-turn signal
section(s) with the through signal message (AGREETH);

4) The presence of multiple through signals (NoTHRU);

5) The presence of a supplemental sign (SIGN);

6) The presence of more than three sections on the left-turn signal
(NoSECTIONS); and, finally,

7) The presence of an arrow on the left-turn signal (ARROW).

All of the above signal face arrangement (SFA) characteristics
require the identification and processing of more than one (or a hard—
to-identify) piece of information present in an SFA in order to arrive
at a decision about the driver's right-of—way privileges associated with
a left turn and may lead to driver confusion especially for older
drivers. A comprehensive list of variables used for intra-interval
comprehension comparisons is presented in table 3.6.

The results of this analysis provide information for the design of

better comprehended signal displays based on knowledge about which

signal characteristics are most likely to lead to better comprehension

56

 

Table 3.6. Signal variables used in comprehension analysis

 

AGREELT
AGREETH

ARROH
ILLUMINATED
NOSECTIONS
NOTHRU

SIGN

Other variables

ARRANGEMENT
CHANGE
COLORS
FLASH
HPOSITION
LFTFLSN

THRUCOL
VPOSITION

 

VARIABLE NAME

Complexity Variables

VARIABLE DESCRIPTION

Concurrence/discordance of simultaneously illuminated left-turn
signal sections

Concurrence/discordance of left-turn and through signal
indications

Presence of left-turn arrow signal section

Number of simultaneously illuminated sections on left-turn signal
Mulber of left-turn signal sections

Number of through signal faces

Sign related to left-turn signal

Stacked three, stacked four, doghouse signal face arrangement
Colors illuminated on left-turn signal during change interval
Left-turn and through signal color configurations

Flashing signal operations configuration

Horizontal signal position (in feet to the left of driver)
Left-turn signal section type used in flashing operations (dark,
arrow, ball)

Through signal color

Vertical signal position

 

of each interval displayed in a particular signal configuration. The

organization of part B of the laboratory data "reanalysis" is shown in

figure 3.5.

 

Figure 3.5. Part B laboratory data reanalysis

 

 

Inter-interval analysis

 

 

Differences among permitted,
protected, red, change and
nighttime displays

 

 

 

 

Intra-interval analyses

 

Red
Change

 

Permitted
Protected

Nighttime

‘Complexity,
Other
Relations with signal
variables
(table 3.6)

 

 

 

 

 

 

57

Hypotheses and statistical methods used for the inter- and intra-interval analysis

Null and alternative hypotheses used to examine comprehension
differences among (and within) stimulus meaning groups presented here
are tested for statistical significance and the results are presented in
the next chapter. Typical testing procedures and statistics are
described below.

Based on the literature review, older drivers are expected to have
higher error rates. However, practitioners need to know whether older
driver comprehension difficulties are significantly different from those
of other drivers and whether such differences are more pronounced during
specific signal intervals. Stimuli categories associated with
significantly higher serious error rates should be addressed
immediately. Typical null and alternative hypotheses for the inter-
interval analysis are:

H: There are no statistically significant differences in
correct answer rates (serious error rates) among stimulus
meaning categories between age groups.

Hg Statistically significant differences in correct answer
rates (serious error rates) among stimulus meaning
categories between age groups exist.

These hypotheses will be tested using a multivariate multi-group
analysis of variance model, appropriate for multiple—response data.
Subjects (considered a random effect) are nested within age groups (AGE)
and crossed with stimulus meaning groups (MEANING), while age and
stimulus meaning are completely crossed. Each stimulus group serious
error rate is examined as a separate dependent variable. Subjects and

stimulus meaning groups are the independent variables. The following

three questions are answered in turn:

1) Does stimulus comprehension vary among age groups;

2) Is stimulus comprehension dependent on stimulus meaning;
and,

3) Does an interaction between comprehension of stimulus

categories and driver age exist.

A summary of typical ANOVA model results is presented in table 3.7.

58

 

Table 3.7. Inter-interval serious error rate comparison

 

 

Source of Variation SS DF MS F Sig of F
Main effects

AGE .26 3 .09 5.65 .001
MEANING 1.47 3 .49 38.31 .000

Z-way Interactions
AGE BY MEANING .13 9 .01 1.11 .354

 

 

 

Comprehension (measured in serious error rate) is statistically
significantly different among driver age groups (F-value 5.65,
significance 0.001). It is also statistically significantly different
among stimulus meaning categories given an F-value of 38.31 and a
significance of 0.000 for variable MEANING. Finally, the F-statistic
value for AGE by MEANING interactions of 1.11 corresponds to a
significance of 0.354 which is not statistically significant, indicating
that comprehension of stimulus group meaning is not statistically
significantly different among age groups. An example of statistics for
age and stimulus meaning groups compiled for each ANOVA model are
presented in table 3.8.

Based on the results presented in table 3.8, the mean serious
error rate for change interval stimuli is .010 (1%). This average is
calculated based on the answers of all subjects (N = 186). Drivers 61
to 75 years of age have an average serious error rate of .014 (1.4%) for
the same stimuli. Older subject results are based on a sample of 42
subjects, and the 95% confidence interval for the older subject average
is a serious error rate between .003 and .026.

Hypotheses tested using similar models are presented in table 3.9.
Following inter-interval hypotheses analysis, intra-interval hypotheses
will be examined using similar statistics. Hypotheses tested in the
intra-interval analysis are presented in table 3.10. A complete

discussion of findings can be found in the next chapter.

 

 

559

 

Table 3.8. Serious error rate statistics for inter-interval analysis

 

Permitted interval stimuli

FACTOR RANGE Mean Std. Dev. N 95 percent Conf.Intvl.
AGE 16 TO 30 .040 .100 48 .011 .070
AGE 31 TO 45 .091 .183 51 .039 .142
AGE 46 TO 60 .082 .119 45 .046 .118
AGE 61 TO 75 .125 .172 42 .071 .179

For entire sample .083 .150 186 .062 .105

 

Red interval stimuli

FACTOR RANGE Mean Std. Dev. N 95 percent Conf.Intrl.
AGE 16 TO 30 .014 .044 48 .001 .027
AGE 31 TO 45 .021 .107 51 -.009 .051
AGE 46 TO 60 .038 .073 45 .016 .060
AGE 61 T0 75 .083 .142 42 .039 .127
For entire sample .037 .100 186 .023 .052

 

Nighttime operations stimuli

 

 

 

 

FACTOR RANGE Mean Std. Dev. N 95 percent Conf.Intvl.
AGE 16 T0 30 .103 .138 48 .063 .143
AGE 31 TO 45 .107 .154 51 .064 .151
AGE 46 TO 60 .144 .143 45 .101 .187
AGE 61 TO 75 .154 .146 42 .109 .200

For entire sample .126 .146 186 .105 .147

Change interval stimuli

FACTOR ' RANGE Mean Std. Dev. N 95 percent Conf.Intvl.
AGE 16 To 30 .011 .035 48 .001 .022
AGE 31 T0 45 .008 .035 51 -.001 .018
AGE 46 To 60 .006 .020 45 .000 .012
AGE 61 T0 75 .014 .036 42 .003 .026

For entire sample .010 .032 186 .005 .015

 

 

Table 3.9. Tested hypotheses, inter-interval comprehension analysis

 

H : There are no statistically significant differences in correct answer rates
(serious error rates) among stimulus meaning categories between age groups.

H,: Statistically significant differences in correct answer rates (serious error
rates) among stimulus meaning categories between age groups exist.

 

 

H : There are no statistically significant differences in correct answer rates
(serious error rates) among stimulus meaning categories for older drivers.

H1: Statistically significant differences in correct answer rates (serious error
rates) among stimulus meaning categories exist for older drivers.

 

 

6C)

 

Table 3.10. Tested hypotheses, intra-interval comprehension

analysis

 

Permitted interval

 

H :

There are no statistically significant differences in correct answer rates
(serious error rates)1 based on:
signal horizontal position (HPOSITION)2
concurrence/discordance of left-turn signal with through signal
(AGREETH)
number of through signals (NoTHRU)
presence of supplemental sign (SIGN)
between age groups (among older drivers)‘.

Statistically significant differences in correct answer rates (serious error
rates) based on the above variables exist between age groups(among older
drivers).

 

Protected interval

 

H

H1:

There are no statistically significant differences in minor error rates3
based on:
signal horizontal position (HPOSITION)
signal vertical position (VPOSITION)
concurrence/discordance of left-turn signal with through signal
(AGREETH)
concurrence/discordance of simultaneously illuminated left-turn
signal sections (AGREELT)
presence of supplemental sign (SIGN)
number of simultaneously illuminated sections on the left-turn
signal (ILLUMINATED)
through signal color (THRUCOL)
between age groups (among older drivers).

Statistically significant differences in minor error rates based on the
above variables exist between age groups(among older drivers).

 

Red interval

 

H :

 

There are no statistically significant differences in serious error rates‘
based on:

signal horizontal position (HPOSITION)

signal vertical position (VPOSITION)

presence of supplemental sign (SIGN)

signal face arrangement (ARRANGEMENT)

presence of a left-turn arrow (ARROH)

between age groups (among older drivers).

Statistically significant differences in serious error rates based on the
above variables exist between age groups(among older drivers).

 

 

 

 

 

 

61

 

 

 

Table 3.10. (cont'd). Tested hypotheses, intra-interval
comprehension analysis

 

Change interval

 

H,: There are no statistically significant differences in correct answer
(serious error rates) based on:
signal horizontal position (HPOSITION)
signal vertical position (VPOSITION)
concurrence/discordance of left-turn signal with through signal
(AGREETH)
concurrence/discordance of simultaneously illuminated left-turn
signal sections (AGREELT)
presence of supplemental sign (SIGN)
number of simultaneously illuminated sections on the left-turn
signal (ILLUMINATED)
through signal color (THRUCOL)
number of through signals (NoTHRU)
between age groups (among older drivers).

H1: Statistically significant differences in correct answer (serious error
rates) based on the above variables exist between age groups(among older
drivers)

 

 

Flashing operations

 

H,: There are no statistically significant differences in correct answer
(serious error rates) based on:
signal horizontal position (HPOSITION)
signal vertical position (VPOSITION)
presence of supplemental sign (SIGN)
through signal color (THRUCOL)
number of through signals (NoTHRU)
left-turn signal indication (LFTFLSH)
left-turn action (FLASH)
between age groups (among older drivers).

 

H1: Statistically significant differences in correct answer (serious error
rates) based on the above variables exist between age groups(among older
drivers)
1. Separate models for correct answer rates and serious error rates, all drivers and
older drivers
2. See table 3.6
3. No serious error rates possible for protected interval stimuli, since drivers have
the right of way
4. No minor error rates possible for red interval stimuli, since drivers do not have the

 

right of way

 

 

62

3.5. FIELD DATABASE

Field database records represent vehicles involved in accidents.
Each record contains vehicle, driver, and accident information as well
as geometric and signal information about the approach on which the
vehicle was moving when it was involved in the accident. The data,
covering a span of 10 years and 324 intersections, were assembled in
separate databases and then combined.

Each geometric file record represents conditions at a given

intersection approach that consists of information about
intersection type, lane assignment and other variables (see
table 3.11). Geometric information was obtained from Michigan
Department of Transportation (MDOT) archives and dates that geometric

changes were enacted were obtained from work contracts.

 

Table 3.11. Geometrics file variables

 

Intersection type: Tee
Four-leg two two-way streets
Four-leg one two-way one one-way street
Number and position of signals (signal profile)
Horizontal and vertical signal position
Lane assignment (geometric profile)
Number of left-turn lanes

Number of opposing-through lanes

Left-turn sign

 

 

 

Signal information was obtained from MDOT intersection files and
controller configuration printouts. Each signal file record corresponds
to an intersection approach and contains variables
pertaining to phasing, signal display complexity, and other

signal-related variables (see table 3.12).

 

63

 

Table 3.12. Signal file variables

 

Approach ID
Signal section arrangement

Controller type: Mechanical
Actuated
Solid State

Left-turn phase type: Permitted
Protected
Protected/permitted

Protected phase type: Lead
Lag

Cycle length in sec.
Permitted duration in sec.
Protected duration in sec.

Number of simultaneously illuminated
left-turn signal lenses

Concurring or discording left-turn
and through signal indications during the:
Red Phase
Yellow Phase
Permitted Phase
Protected Phase
Flashing operations mode

Major or minor approach

 

 

 

Multiple records were coded in the corresponding files for
intersection approaches where changes were made to geometric and/or
signal approach parameters; beginning and ending dates (defining the
period during which a particular configuration was in effect for a
particular approach) were coded for each record. Only intersection
approaches with a left-turn movement facing an opposing through movement
were coded in the two files: for example, there is only one leg with
left-turns facing an opposing through movement present at
T-intersections, the one allowing left-turns off the main direction onto
the Tee leg. Similarly, at intersections of one-way with two-way

streets, there is only one approach for which left-turning drivers face

 

64

opposing traffic.

Accident records were obtained from accident records maintained by
MDOT. All accidents within 150 feet from each study intersection were
selected based on milepoint information coded in the accident record.
Each original accident record containing information about all vehicles
involved in an accident was separated into vehicle records, one for each
vehicle involved in the accident (accident-related variables contained
in the database are shown in table 3.13). Vehicle direction before the

accident, from the original accident record, was used to assign vehicles

to intersection legs.

 

Table 3.13. Accident variables

 

Hour, month and year
accident occurred

Weather, light, and road surface,
conditions

Injury severity
Accident type
Vehicle type
Driver age
Driver residence
sex
violation

Contributing circumstances

Visual obstructions

 

 

 

In all, 3004 signal, 1217 geometric, and 187,715 vehicle
(accident) records were entered in the corresponding files. Geometric
and signal information records were merged into an intersection
configuration file in such a manner that each new record represents a
time period during which the approach signal and geometric configuration
remained unchanged. The intersection configuration file contains a
total of 3444 records. Finally, each vehicle record was merged with the

appropriate intersection configuration record according to where (i.e.,

 

 

 

65

 

which intersection leg) and when an accident occurred. This final merge
produced the field database (figure 3.6). The field database allows the
analysis of the effects that particular signal and geometric
configurations have on the occurrence of an accident. Information at
the intersection approach level is necessary because in many instances

different approaches of the same intersection use different left-turn

 

signal phasing and have dissimilar geometric configurations.

 

Figure 3.6. Field database

 

 

 

 

 

Geometric
data Signal data Vehicle data
table 3.11 table 3.12 table 3.13

 

 

 

 

 

 

 

 

 

Merge

l

Intersection

 

 

 

 

 

 

configuration /
data
Merge
Field
database

 

 

 

 

 

 

3.6. FIELD DATA ANALYSIS

Based on the literature review, it is expected that as drivers
age, they have a higher tendency to become involved in accidents.
Furthermore, older driver propensity for higher accident involvement is
expected to be more pronounced for more complex intersection maneuvers.
The purpose of the field data analysis is to answer a number of
questions:

1) Does accident involvement increase with increasing age;

2) Is older driver accident (over)involvement related to
intersection maneuver (driving straight through the

66

intersection versus turning left or right);

 

3) Is older driver accident (over)involvement present during
both normal (daytime) and emergency/ nighttime signal
operations;

4) Is older driver accident (over)involvement related to left-
turn signal control type;

5) Which approach geometric and signal variables are associated
with higher left-turn accident involvement, especially for
older drivers; and, finally,

6) What are the most prominent accident characteristics
differences that set older drivers apart from other driver

age groups.

 

The field data analysis closely follows the organization of
laboratory data analysis in order to facilitate comparisons between
findings from the two efforts. Thus, for example, accidents during
daytime and nighttime left-turn signal operations are examined
separately, also accidents at approaches with permitted, protected, and
protected/permitted operations are examined individually, closely
following the laboratory data analysis format. Based on signal
controller information daytime is defined as the period between 7:00am
and 10:59pm and nighttime as the remainder of a 24-hour period.
Intersections with 24-hour full-color operations were excluded from
nighttime statistics to avoid the introduction of bias due to
differential older driver daytime/nighttime performance at full color
locations. The abundance of accident information allows the
introduction of finer (i.e., narrower) driver age groups than those used
in the laboratory data analysis. The original age group limits are
augmented depending on the amount of information available. Thus,
comparisons between the two analyses are still possible, but a more
refined pattern of changes in accident statistics is obtained. However,
throughout the field data analysis driver age groups identical to those
used for the laboratory data are maintained allowing direct comparisons
between the two analyses.

A summary of intersection configuration information can be found
in table 3.14. Field database analysis is organized into four parts as
outlined in figure 3.7, proceeding from general descriptive statistics

analyzing relations of driver age with intersection maneuver to an

 

 

 

 

67

 

Table 3.14. Intersection configuration information

 

GENERAL INFORMATION

Number of records

 

INTERSECTION CONFIGURATION FILE HATCHED RECORDS

 

 

 

 

 

 

 

 

Both geometric and signal information available 2409
Missing geometric record (41
Missing signal record 594
APPROACH IS MAJOR/MINOR
Major approach 1240
Minor approach 1182
INTERSECTION CONFIGURATION
Tee 173
4-leg two two-way streets 2681
4-leg two-way & one-way 147
SIGNAL PROFILE (most common configurations)
1 Left turn 1 Through & right-turn 1653
1 Left turn 1 Through 170
1 Left turn 1 Through 1 Through 8 right-turn 872
2 Left turn 1 Through & right-turn 116
LEFT TURN SIGNAL HORIZONTAL POSITION (feet left of driver)
up to 3 feet 1466
4-16 feet 1759
17-38 feet 219
CONTROLLER TYPE
Mechanical 2194
Actuated 224
Solid State 432
LT PHASE TYPE
Permitted Green Ball 1864
Protected Green Arrow 400
Protected Green Arrow + Green Ball 40
Protected/Permitted Flashing Red ball, Green Arrow 337
Protected/Permitted Green Ball + Green Arrow 152
CYCLE LENGTH
45-60 sec. 918
65-75 sec. 726
80 sec. 875
90-120 sec. 224

 

 

 

 

 

 

 

6E3

 

Table 3.14. (cont'd).

 

PERMITTED PHASE

Number of records

 

DURATION IN SECONDS

up to 21 sec.
22-27 sec.
28-35 sec.
36-66 sec.

772
549
467
521

 

PROTECTED PHASE

Number of records

 

 

PHASE TYPE

Lead 211
Lag 555
DURATION IN SECONDS

up to 6 sec 256
11-52 sec 334
7-10 sec 226

 

FLASHING OPERATIONS

Number of records

 

 

Flashing yellow
Flashing red

Dark

24-hour operation

 

LEFT-TURN THROUGH
SIGNAL SIGNAL
706 894
1046 910

56
1032 1032

 

 

 

69

 

Figure 3.7. Field data analysis flowchart

 

 

 

General descriptive statistics for left-turn, through, and
right turn intersection accidents-relations with driver
age. All approaches:

Full color hours

Flashing hours

With permitted left-turn phasing

With protected left-turn phasing

With permitted/protected left-turn phasing

 

 

 

 

Involvement ratios for
All accidents full color operation hours
All accidents flashing operation hours
Left-turn accidents full color operation hours
Through accidents full color operation hours
Right-turn accidents full color operation hours

 

 

 

i

 

Involvement ratios for left-turn accidents
Full color operation hours:
All approaches
Approaches with permitted phasing
Approaches with protected phasing
Approaches with permitted/protected phasing

 

 

 

Flashing operation hours:

 

All approaches
Left—turn accident involvement relations with:
Signal control characteristics
Signal face and phase type
Approach geometry and left-turn signal position
Accident and driver characteristics

 

 

 

 

 

 

7O

 

examination of involvement ratios for particular intersection maneuvers
(e.g., left turn, through, right turn) and further focusing the analysis
into a detailed examination of left-turn accident involvement ratios for
each left-turn phasing strategy present in the database (permitted,
protected, permitted/ protected). The analysis concludes with an

extensive examination of relations of signal control, approach geometry,

 

and driver and accident characteristics, with older driver left-turn
intersection accident experience.

Statistical method used for field database analysis

In view of the lack of a measure of exposure that is not age-

 

biased, the driver-at-fault/innocent victim ratio method (D1/D2)
discussed in detail in the literature search is used in the present
analysis, since it provides a differential measure of exposure based on
driver age. Over- (under-) involvement of each driver age group is
based on whether more (less) drivers within a particular age group tend
to be at-fault than innocent victims in multi-vehicle accidents.

Based on information presented in table 3.15, the involvement

 

 

 

 

 

 

 

 

Table 3.15. Male driver D1/D2 table for daytime left—turn accidents
7 at intersections with permitted left-turn phasing
Driver not at fault (D2) age
15-19 20-45 46-60 60+ Row
total
15-19 91 293 66 41 491
Driver 20-45 188 800 123 87 1198
at fault
(D1) age 46-60 56 230 33 20 339
60+ 65 296 52 38 451
Column 400 1619 274 186
total

 

 

 

 

 

 

 

 

 

ratio of 15—19 year-old male drivers involved in daytime left-turn

accidents at intersections with permitted left-turn phasing is:

71

fl = —491 = 1.227

D2 400

Similarly, the involvement ratio for drivers 20-45 is 0.74; for
drivers 46-60, 1.237; and for drivers over 60, 2.42. These involvement
ratios indicate that among male drivers, only those between the ages of
20 and 45 are underinvolved in this particular type of accidents.
Drivers from this age group are at fault on 1198 occasions, while they
are innocent victims on 1619 occasions out of a total of 2817 occasions
when a driver from this age group was involved in an accident. There
were 800 accidents where both involved drivers were males between ages
20 and 45. The total number of accidents for this gender and age group
is 1619 + 1198 - 800 = 2017. All other male driver age groups were
overinvolved in accidents (i.e., the involvement ratios are greater
than 1), with drivers over age 60 showing the most pronounced

overinvolvement at a ratio of 2.42.

 

 

CHAPTER 4

ANALYSIS OF LEFT-TURN SIGNAL COMPREHENSION IN THE LABORATORY
4.1. INTRODUCTION

The material that follows is organized into three major parts as
outlined in the analysis methodology (figure 3.4):

I) General subject demographic information;
II) Answer correctness relations with:
confidence in given answer,
subject sex,
miles driven per year, and
driving experience; and,

III) Answer correctness relations with various signal

characteristics.

Demographic information (part I above) is analyzed with general
descriptive statistics and analysis of variance (ANOVA) models
(summarized in appendix B). Comprehension (parts II and III above) is
evaluated using two criteria, namely the overall and older driver error
and correct answer rates. The comprehension analysis is organized into
paragraphs following similar outlines with an introduction, a
presentation of error rates and correct answer rates for all and older
drivers and a section summary.

4.2. SUBJECT CHARACTERISTICS ANALYSIS-GENERAL COMPREHENSION ANALYSIS

Of the 191 subjects in the laboratory experiment, fifty-six
percent (107) were males. Subjects were distributed fairly evenly over
the four age groups: ages 16-30 (50 subjects), 31-45 (51), 46—60 (48),
>60 (42) so that enough individuals are present in each group for
statistical inferences (see table 3.1.). A graphic summary of various
demographic and driving characteristics of the subjects is presented in
figures 4.1-4.3. About 86% of the respondents in the youngest and
oldest age groups drive less than 15,000 miles per year compared to
about 72% for the other two age groups. About 62% of the youngest
drivers drive less than 10,000 miles per year. This percentage drops to
44% for the 31-45 year-olds, and then increases until it becomes 75% for

the oldest driver group (figure 4.1). The youngest and oldest drivers

tend to drive less than the other two age cohorts with the oldest group

72

 

 

 

73

 

 

Figure 4.1. Miles driven annually by age

 

 

IIIIIIIIIII
IIIIIIIIIIII

      

\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\

 

       

"l'...".'..'l'-'.'l"\.

Id

IIIIIIIIIIIIII
...-IIIIIIIII

\\\\\\\\\\\\\m

 

60

50+
40+
0
20+
10~
o

Eoocoa

60+ years old

31 -45 years old

Driver age

<7k -7k-10k -11k-15kfl >15k

 

 

 

 

74

 

 

Figure 4.2. Miles driven annually by age (male drivers)

 

 

§§
,.\\\x\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\§

 

 

 

\\\\\\\\.\\\\\\\\\\\\\\\\\ \\\\\\\\\\\\\\\\\._
\\\\\\\\\\\\\\\\

46-60\ years old

 

are old

\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\_
\\\\\\\\\\\\\\\\\\\\\\\\\\\A

16-30;\

 

 

# 0 a 4 4 Al 4 J1 d
0 5 w 5 0 5 0 5 0
4 3 2 2 1 1

Eoocoa

60+ years old

31 -45 years old

Driver age

- 7k-10k11k-15k E > 15k

-<7k

 

 

 

75

 

 

Figure 4.3. Miles driven annually by age (female drivers)

 

 

 

 

 

80

70+

60+

50+
0

5026..

30.4
20-
1 0-

46-60 years old
60+ years old

31 —45 years old

old

16-30 years

Driver age

- 7k-10k11k-15k E > 15k

-<7k

 

 

 

 

 

76

driving less than any other.

When mileage is further analyzed by sex (figures 4.2 and 4.3), an
important difference between male and female subjects becomes evident: a
larger percentage of female drivers tend to drive less than 10,000 miles
per year for all age groups. This difference is especially striking for
the 46-60 year-old drivers 71% of female drivers drive less than 10,000
miles per year compared to 22% of male drivers.

Male drivers drive more miles per year as they age, until age 60.
Males older than 60 show a dramatic decrease in miles driven as they
drive even less than the youngest drivers. A similar trend was observed
for female drivers, except there is a drop in miles driven for the 46~60
year- old drivers and this trend continues through the oldest driver
group where 76% drive less than 7,000 miles per year. The percent of
female drivers that drive 15,000 miles or less annually remains around
87% for all age groups while for males this percentage represents only
the youngest and oldest drivers. About 65% of the other male age groups
are within this range. Analysis of variance (ANOVA) models were
developed to analyze miles driven per year (dependent variable) by
subject age and sex. Using AGE as the only independent variable,
statistically significant differences in miles driven per year were
found (an = 2.62, FProb = .053). A model including both sex and age as
independent variables, showed statistically significant differences for
main effects (an = 3.95, FProb = .004), while the age—sex two-way
interaction was not found to be statistically significant. A multiple
classification analysis shows a sample average of approximately 12,000
miles driven annually with the youngest and oldest drivers driving less
than the average (-650 and —4700 miles per annum--adjusted for other
independent variables-- respectively), and drivers 31 to 45 and 46 to 60
years old driving above the average (+1800 and +3100 miles per annum
respectively).

Relations of answer correctness to estimated miles driven, driver

age, years of driving experience, and sex are presented in figures

77

4.4-4.8. For both sexes, the percent correct responses among confident
answers (confidence score of five) is higher than the percent correct
responses among less confident answers (approximately 72% for confident
responses compared to an average of about 46% correct answers for less
than confident responses--see figure 4.4). However, the number of
correct plus minor error responses tends to be reasonably constant among
levels of confidence (about 94%). This is an indication that, when
drivers are not certain about the correct action, they tend to behave in
a conservative manner since both drivers who are confident in their
responses and those who are not tend to commit a relatively small number
of serious errors (about 6%). Answer correctness at various levels of
answer confidence is similar for both sexes.

For both sexes, the percent of correct answers increases from
about 57% for under 7,000 miles driven annually to about 71% for
11-15,000 miles driven annually and drops to about 62% for over 15,000
miles driven annually. Correct plus minor error answers are about 95%
regardless of miles driven (figure 4.5).

Two ANOVA models using annual mileage driven as the independent
variable and correct and serious answer rates as the dependent variables
respectively, showed the following: correct answer rates differ between
annual miles driven categories (an = 5.99, FProb = .001) while no
significant differences were found for serious error rates (an = 0.86,
FProb = .462). The highest percentage of correct answers, and at the
same time lowest rates of minor and serious errors, were attained by
individuals driving 10-15,000 miles annually.

The percent of correct answers decreases with age in a similar
fashion for both sexes: the youngest group correctly answered about 70%
of the questions, and this percentage declines steadily to the oldest
group which correctly answered only 51%. When the correct plus minor
error rate is considered, the differences between age groups follow the
same trend but are less pronounced--the range is from 96% to 93%. The
oldest group appears to make up for lack of comprehension by acting in a

conservative manner-e.g., giving the right-of—way to other drivers

78

 

 

 

 

     

V//////////

.IHMNIII”IIIIIIIIIIINIIII
.mmmmmmmmmmmmmmmmmmmmmmmm

§§
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\R

   

 

 

 

 

IHQHHHHHHH HFT FFL

 

 

HHHHLHHH~_~fl----HHHHH
_

 

 

 

 

 

80

Figure 4.4. Answer correctness by confidence in given answer

70-l

60+

11 1
O 0 0
5 3

«coocoa

20+
10
0

Serious error

Minor error
Correctness level

Confidence 3

Least confident - Confidence 2

E Confidence 4

Most confident

 

 

79

 

 

Figure 4.5. Answer correctness by miles driven per year

 

 

1...
.m
.w

rror
Correctness level

9
r.

 

 

 

 

so
704
60-i
50-1

10'i
0

< 7k - 7k-10k11k-15k m >15k

 

 

80

 

 

Figure 4.6. Answer correctness by driver age

 

 

lllll

IIIII

IIIII

IIIII
\

 

Serious error

§

\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\m
§§a

 

 

80

70%
60-l

A a 4
O 0 O
5 4 3

=5qu

20%
1 O
0

Minor error
Correctness level

16-30 years old - 31-45 years old 46—60 years old fl 60+ years old

 

 

81

 

 

Figure 4.7. Answer correctness by years driving experience

 

 

 

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

§

 

.\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\m
s\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\a

 

AI Iql * + + N
c 0 O O o o o 0
7 6 5 4 3 2 1|
«coucoa

Minor error
Correctness level

0-12 years - 13-24 years 25-38 years m 38+ years

 

 

 

 

 

 

 

 

 

 

83

(figure 4.6). Drivers with less than 24 years of driving experience
have a correct answer rate of approximately 68 percent. From that point
on there is a continuous decrease in the number of correct answers with
increasing driving experience until it becomes 52% for drivers with more
than 38 years of driving experience (figure 4.7). Although figure 4.7
does not make intuitive sense by itself, it is in line with the
discussion about driver age, since the more experienced the drivers, the
older they are (in general). It also implies that experience may not
overcome other problems for older drivers.

Significant differences among driver experience groups were found
for correct answer rate (an = 8.74, FProb = .000) as well as a serious
error rate ANOVA model (an = 3.08, and FProb = .029).

It has been observed that the causes for over-representation in
accidents are different for the youngest and oldest drivers. Younger
drivers tend to be cited for driving too fast and, although they have
better comprehension and reaction times, they tend to overestimate their
capabilities and those of the vehicles they drive. Older drivers tend
to be conservative but have problems with stimulus comprehension and
reaction times (73, 74, 126). Figures 4.6 and 4.7 suggest that signal
comprehension might be one area where the youngest and oldest drivers
differ--the oldest drivers show a much lower rate of correct
interpretation of left-turn stimuli than their younger counterparts.

Female subjects had about 58% correct answers while their male
counterparts had about 65% correct answers. Again, when correct and
minor error answers are combined, the sexes are tied at 95% (figure
4.8). While correct answer rates were found to differ among sexes (an
= 10.75, FProb = .001), serious error rates are essentially identical
(an = 1.07, FProb = .302) at seven percent.

Other than the differences mentioned above, none of the data
presented in figures 4.4-4.8 had any significant differences among

sexes o

84

Summary of subject characteristics analysis findings

Youngest and oldest drivers drive less miles per year than other
drivers. Female drivers drive less than their male counterparts with the
most pronounced differences being among the older than 45 year-old
drivers. Drivers of both sexes drive more miles until a certain age
cutoff and then drive less (even less than the youngest drivers). The
reduction in miles driven with progressing age occurs for 46-60 year-old
female drivers and continues through the oldest female drivers, while
for male drivers it is only evident for the oldest age group.

As would be expected (and hoped), a significantly higher rate of
correct answers is found among confident answers (with a confidence
score of five) than among less than confident answers. Cumulative
correct and minor error answers are at an almost constant level
indicating that when drivers are not certain about the meaning of a
display, they choose to surrender their right-of—way.

The rate of correct responses increases with annual miles driven up
to about 15,000 miles/year and then declines. Statistically significant
differences in correct answer rates among levels of miles driven
annually were verified using ANOVA; no such differences were found for
serious error rates.

The youngest drivers have the highest rate of correct answers and
the oldest drivers have the lowest, with correct answer rates steadily
declining with age. The cumulative correct and minor error rates follow
the same trends although with a much less pronounced decline in the
differences with progressing age.

The rate of correct answers steadily declines with increasing driver
experience, a trend associated with driver age. A much less pronounced
increase in serious error rates is present across driving experience
levels. However, ANOVA models for both correct and serious error rates
show statistically significant differences in the distribution of answer
correctness across driver experience groups. Individuals with up to 24
years driving experience have the best comprehension rates.

Statistically significant correct answer rate differences were found

85

among sexes with males having a significantly higher rate of correct
answers. However, no comprehension differences in serious error
responses were found among sexes.

4.3. ANALYSIS OF ANSWER CORRECTNESS RELATIONS WITH VARIOUS SIGNAL
CHARACTERISTICS

The analysis of relationships between signal characteristics and
answer correctness consists of two components (see chapter 3), the
inter- and intra-interval comprehension comparisons (figure 3.5).
Inter-interval comparisons were used to examine the existence of
comprehension differences across age groups based on interval type.
Intra—interval comparisons were used to identify the existence of
comprehension differences across age groups among stimuli used for the
same interval type. Comprehension effects of individual signal
characteristics as well as interactions among such characteristics were
investigated.

Inter-interval analysis

Four ANOVA models are used to address the inter-interval analysis
hypotheses presented in table 3.9. In the first model, serious error
rate is used as the independent variable to examine driver comprehension
of interval meaning by driver age. Included in the database are all
subject responses to permitted, change, and red intervals as well as
flashing operations stimuli. A second model, restricted to older
subjects, uses the same dependent variable and interval meaning as an
independent variable. Responses to stimuli used for the protected phase
are not included in these models since no serious errors are possible
when drivers have the right-of—way. The models are used to derive
conclusions about which phases are least understood by drivers to the
extent that they may violate other drivers' right-of—way (i.e., commit
serious errors) and potentially become involved in serious accidents.

Finally, two ANOVA models, similar to the ones mentioned above
(i.e., one for all subjects and one for older subjects) are constructed
to examine correct answer rates as the dependent variables. Protected

stimuli are included in the database. Intervals with low correct answer

86

rates need particular attention by traffic engineers; those with high
correct answer rates may be given preference. The results of all four
ANOVA models are summarized at the conclusion of this section.
Serious errors-all drivers

Responses indicating that the driver would violate the right of way
of other drivers are classified as "serious errors." Serious errors
are, therefore, not possible for stimuli indicating a protected
left-turn phase.

When serious error rates (regardless of stimulus meaning) are
compared across age groups using a one-way ANOVA, serious error rate
differences are detected between driver age groups (an = 3.15, FProb =
.026). (Note that statistical details are provided in appendix B).
Figure 4.9 shows 95% confidence intervals by driver age for serious
error rates. There is a statistically significant increase in percent
serious errors with increasing age (from 5.3% for the youngest drivers
to 9.4% for the oldest). The two youngest driver groups have small
comprehension differences, but differences increase in an accelerated
fashion as age progresses.

Comprehension scores examined under a multivariate analysis of
variance model with repeated measures applied to multiple groups show
statistically significant comprehension differences among stimulus
meaning categories (an = 38.31, FProb = .000) across age groups (an =
5.65, Ff,ob = .001-—see appendix B). No significant age-stimulus meaning
interaction effect was detected (an = 0.01, FProb = .354). The 95%
confidence intervals of responses to each stimulus group are shown in
figure 4.10. Best understood are stimuli used for the change interval
with an average serious error rate of only 1.0%. Next are red interval
stimuli (3.7%) and permitted stimuli (8.3%). Least well understood are
flashing mode stimuli with an average of 12.6% serious error rate. The
95% confidence intervals for any two stimuli groups do not overlap, but
the upper limit of permitted stimuli coincides with the lower limit of

flashing mode stimuli.

87

 

 

Serious error rates

Figure 4.9. All drivers all stimuli serious errors

95% confidence intervals for the means

 

11"

10"

8" _

 

7.4

6%

 

5-4

 

4a

 

 

(a)

 

 

16-30 TYrs old 31 -45 er3 old 46-60 Yrs old 60+ Yrs old
Driver age

 

 

88

 

 

Serious error rate

Figure 4.10. All drivers all stimuli serious error rates

95% confidence intervals for the means

 

141

12d

10‘

8-1

6-4

4...

 

 

PemTlitted Red interval Flashing orange
Stimulus meaning

 

 

 

89

Serious errors-older drivers

A multivariate analysis of variance model with repeated measures
applied to older drivers shows statistically significant differences
among stimulus meaning groups (an = 9.73, FPM = .000). The rank of
stimulus meaning groups based on stimulus comprehension score remains
the same as for the whole driver population as shown in figure 4.11.
However, serious error rates are magnified among older drivers. Best
understood are stimuli used for the change interval with an average
serious error rate of 1.4% followed by red interval stimuli with an
average serious error rate of 8.3%, the permitted interval at 12.5%,
and, finally, flashing mode stimuli at 15.4%.
Summary and interpretation-serious errors

Stimuli demanding the most "obvious" driver reactions (such as
preparing to stop for a change interval or red ball) have the highest
comprehension scores and those involving some interpretation because of
a special meaning pertaining to the left—turning maneuver have lower
scores. For example, flashing mode stimuli require more interpretation
because right-of—way rules change depending on the combination of
flashing colors on the left and through signals--red with red have a
different meaning than red with yellow.

Comprehension differences based on stimulus meaning are similar
across all age groups (no significant stimulus meaning-driver age
interaction effects were identified). For example, change interval
stimuli are best and flashing stimuli least well comprehended for all
age groups. However, differences among stimuli groups are more
pronounced among older drivers.

Signal complexity may explain both stimulus meaning and age group
patterns of serious error rates as demonstrated in the following
examples. The most complex situation is that of a flashing red ball
where stimulus meaning varies according to the through signal
indication. Although the driver always has to stop, a through flashing
yellow ball requires the driver proceed if there is no opposing traffic;

a flashing red ball requires the driver to proceed if there is no

9O

opposing or cross street traffic. The complexity of this situation is
clearly reflected by the high serious error rates for such stimuli,
especially for older drivers. A steady yellow or red ball, the simplest
displays with a singular meaning, i.e., "prepare to stop" or "stop-you
do not have the right of way," is associated with the lowest serious
error rates both for all driver ages.
Correct answers-all drivers

Comprehension scores examined under a multivariate analysis of
variance model with repeated measures applied to multiple groups shows
differences among age groups (an = 13.37, FProb = .000) and stimulus
meaning categories (an = 180.32, FProb = .000). A significant age-
stimulus meaning interaction effect was also detected (Pam = 2.42, FProb
= .004). Here, again, driver comprehension is deteriorating with age
(figure 4.6). The ranking of stimuli groups (best comprehended group
has a rank of one) is different than that based on serious error rates,
the differences being due to the presence of differing "minor error"
rates among stimuli groups. Best comprehended are red interval stimuli
followed by change, protected, permitted, and flashing operations
stimuli.
Correct answers-older drivers

Comprehension varies across stimulus meaning categories in terms
of correct answer responses among older drivers (an = 47.38, FProb =
.000). The ranks of stimulus meaning groups based on stimulus
comprehension scores are identical to those for all drivers. The 95%
confidence intervals of responses to each stimulus group are shown in
figure 4.12. Best understood are stimuli used for the red interval with
an average correct answer rate of 92%, followed by change interval
stimuli (63%), protected (49%), permitted (38%), and finally, flashing
operations (32%).
Summary and interpretation-correct answers

Statistically significant stimulus comprehension differences exist
across stimulus meaning and driver age groups. Best comprehended are

stimuli used for the red phase interval, followed by those used for the

91

 

 

Serious error rate

Figure 4.11. Older drivers all stimuli serious error rates

95% confidence intervals for the means

 

20

18-

16-

14~

 

 

 

 

 

Permitted Red interval Flashing ChaInge
Stimulus meaning

 

 

 

92

 

 

Serious error rate

Figure 4.12. Older drivers all stimuli correct answer rates

95% confidence intervals for the means

 

100

801

70-i

60-l

50-1

40-l

3O-i

 

20

 

T

Permitted Protected Red F lashing Chalnge
Stimulus meaning

 

 

 

93

change, protected, permitted, and flashing intervals respectively. The
rank orders of stimulus meaning groups based on correct answer rates are
identical for all and older drivers. Stimulus complexity may explain
comprehension differences, since simple stimuli (e.g., red, change
interval stimuli), with singular meanings, are better comprehended than
more complex stimuli (e.g., flashing operations stimuli).

Inter-interval summary

Stimulus comprehension measured both in serious error and correct
answer rates has been found to depend on stimulus message, and driver
age. It is important to note that stimulus meaning ranks based on the
same type of comprehension scores (i.e., serious error, correct answer
rates) are identical for all and older drivers. Best comprehended are
the simple unambiguous stimuli used for the change and red intervals,
while least well understood are those used for flashing operations.
Comprehension was found to deteriorate to a statistically
significantly degree with advancing age. Serious error rates are the
highest and correct answer rates the lowest for drivers over 60 years of
age.
Infra-interval analysis
The intra-interval comprehension analysis includes separate

discussions on the permitted, protected, red and change intervals, and,
finally, flashing operations. Each paragraph begins with a discussion
of driver comprehension explanatory variables that were analyzed
relevant to the particular set of stimuli and follows a format similar
to the one used in the inter-interval analysis. Specifically, ANOVA
models using serious error rate (minor error rate for stimuli used in
protected phasing where serious errors are not possible) as the
dependent variable and driver age as well as signal position,
arrangement, and various complexity variables as independent variables
are presented for all and older drivers, followed by similar models
using correct answer rate as the dependent variable. A summary of
findings concludes each section with an overview of the intra-interval

analysis following the discussion on flashing operations.

94

Permitted stimuli

Eleven stimuli used for the permitted interval are included in
this analysis. Independent variables used in the ANOVA models include
subject age, signal horizontal position, concurrence or discordance of
left and through signal indications, and number of through signals:

i) Signal horizontal position (HPOSITION). A signal placed at the
extension of the left-turn lane may lead the driver to different
reactions than a signal placed in a "shared" position, between the
left-turn and the through lanes. Since the green ball has a different
meaning for a left-turning driver (yield to opposing traffic) and a
driver moving straight-ahead (you have the right-of-way), signal
positioning may be crucial in reinforcing the notion that the driver
does not have the right-of-way. It is reasonable to hypothesize that the
closer the left-turn signal is to the through signal, the higher the
potential for driver misinterpretation, and the further removed from the
through signal, the higher the possibility that the driver will
correctly interpret the permitted meaning.

ii) The concurrence or discordance of the left-turn signal
indication with that of the through signal indication (AGREETH). A
concurring indication may lead the drivers to believe they have the
right-of—way, since the display itself does not differentiate between
right-of—way rules applying to left-turning (proceed if there is no
opposing traffic) and through drivers (proceed, you have the
right-of—way): proper driver behavior depends absolutely on driver
ability to interpret signal meaning relative to his/her movement
intentions. On the other hand, a discording indication may make drivers
more cautious, since signal faces indicate that through and left-turning
movements have distinct right—of—way obligations. It should be kept in
mind that only special displays (such as flashing yellow or red ball on
the left-turn signal) use discording left-turn messages. Since the
through signal is always a green ball, all "conventional" displays using
a green ball on the left-turn signal can only be concurring.

iii) The number of through signals (NoTHRU). The presence of two

95

through green balls may affect driver interpretation of a permitted
left-turn compared to stimuli using only one through signal. For
example, the presence of multiple signal heads increases stimulus
complexity and might adversely affect drivers vulnerable to decision-
making problems (e.g., older drivers).
Permitted: serious errors-all drivers

Serious errors during the permitted phase are associated with
responses indicating that drivers incorrectly thought they had the
right-of—way while turning. The ANOVA results presented in the
following paragraphs use serious error rates as the dependent variable.

Serious error rates are not statistically significantly different
across subject age levels (an = 0.654, FProb = 0.582) (see appendix B)
while they are significantly different depending on whether the left-
turn signal indication is concurring or discording with the through
signal indication (an = 43.73, FProb = 0.000). Concurring left-turn and
through signal indications are, as expected, associated with higher
serious error rates (13.2%) than discording (1.4%).

No significant comprehension differences were found between
stimuli using one and two through signal heads. However, comprehension
is significantly different for locations with left-turn signals mounted
between the left-turn and through lanes and those where signals are
placed along the left-turn lane centerline. Between-lane placement is
associated with a lower comprehension rate (12.8%) than the straight-
ahead position (6.5%). This result is consistent with the hypothesis
that a closer placement to the through signal may lead drivers to
mistakenly assume they have the right of way.

Permitted: serious errors-older drivers

Concurrence or discordance of the left-turn signal with the through
signal (AGREETH) was found to be related to older driver comprehension
(an = 15.23, FProb = 0.000). As with all drivers, concurring left-turn
and through signal indications are associated with higher serious error
rates (17.8%) than discording (1.8%), following the same pattern (only

exaggerated) identified for the entire subject sample. The number of

96

through signals (NoTHRU) and left-turn signal horizontal position
(HPOSIT) were not found to be associated with serious error rates for
older drivers.

Permitted: summary and interpretation-serious errors

The meaning of a green ball on the left-turn signal needs driver
interpretation to distinguish it from a green ball addressed to the
through driver. A permitted phase green ball is a "passive" signal in
the sense that it does not actively relate its distinct message to the
driver. Three techniques to convert such a "passive" signal into an
“active," one that will provide additional information to the driver
about its special message were examined here: use of discordant left—
turn and through signal displays, positioning of the signal, and number
of through signals. (Not enough data were available for reliable
statistical inferences on the presence of a supplemental sign).

It is reasonable to assume that situations where the left-turn and
through signal meanings are kept distinct will be associated with higher
driver comprehension scores. Thus, wider horizontal separation of left—
turn and through signals, one instead of two through signal heads, and
discording left-turn and through signals were expected to be associated
with lower serious error rates. This was found to be the case.

Left-turn signal concurrence or discordance with the through signal
message is associated with serious error rates in the same manner for
older drivers as for the whole population, only the effect is more
pronounced: concurring messages have a detrimental effect on driver
comprehension and discording messages have a beneficial effect.

Signal horizontal position impacts driver comprehension when the
entire sample was considered. They were not significant in explaining
older driver comprehension. Signal placement between the left and
through lanes versus along the left—turn lane centerline was found to be
associated with lower driver comprehension scores.

However, subject age was not found to contribute to serious error

rate differences among subjects.

97

Permitted: correct answers-all drivers

The only correct answer during the permitted phase is "turn left
without stopping, unless you have to yield to opposing traffic."
Correct answer rates were found to be significantly different among age
groups (an = 7.545, FProb = 0.000). Moreover, comprehension
deteriorates with advancing age with the youngest drivers interpreting
66% of the permitted stimuli correctly and drivers over age 60 correctly
interpreting only 38% of the stimuli.

No significant comprehension differences were identified using ANOVA
models including age and: concurrence/discordance with the through
signal; number of through signals; or signal horizontal placement. No
variable other than subject age was found to be statistically
significant, and all interactions with age were found to be non-
significant.

Permitted: correct answers-older drivers
No statistically significant comprehension differences were

identified for concurrence/discordance with the through signal; number
of through signals; or left-turn signal horizontal placement.
Permitted: correct answers-summary

Correct answer rates deteriorate with increasing age but no other
variables were found to be statistically significant at the .05
significance level.
Permitted: summary and discussion

Permitted phase stimuli have the lowest correct answer rates
(average 38%) and highest serious error rates (average 8.2%—see figure
4.10) among stimuli used for normal, non-flashing operations. This
finding is especially important in light of the popularity of permitted
phasing and the fact that the permitted phase commonly uses a large
fraction of intersection green time thus providing ample opportunity for
a large number of violations of right-of—way rules by drivers that
misinterpret its meaning.

Concurring left-turn and through signal indications had worse

performance than discording signals, while one through signal was

98

associated with better than average comprehension, compared to two
through signals. Signals placed in the straight-ahead position had
better than average performance compared to those placed between lanes.
Serious error rates were found to increase with advancing age, but not
to a statistically significant degree. No variables, other than subject
age, were found to be related to correct answer rates.

Results associating serious error rates with concurrence or
discordance of left-turn with through signal displays, number of through
signals, and signal horizontal position can be easily translated into
practical measures to improve permitted phase comprehension in the
field. For example, the typical Michigan display with discording left-
turn (flashing red ball) and through (green ball) indications placed in
the straight-ahead position can be expected, based on the preceding
analysis, to be better comprehended than a typical stacked three display
(red, yellow, and green balls--green ball illuminated) mounted between
the left-turn and through lanes. Since the same variables affect all
drivers regardless of age, and in the same direction, it is expected
that engineering measures designed to benefit older drivers, will
benefit all drivers as well.

Protected stimuli

Driver comprehension analysis of the twenty stimuli used for the
protected interval of minor error rate as the dependent variable. Only
correct answers and minor errors are possible for protected phase
stimuli, since subjects have the right of way they cannot violate
another driver’s right of way.

In addition to subject age, independent variables examined in the
present section include the presence of a left-turn supplemental sign,
and the configuration of illuminated lenses on the left-turn and through
signals:

I) Sign presence (SIGN). The presence of left-turn
supplemental signs may enhance driver understanding by
clarifying the left-turn signal display message. On the
other hand, sign presence increases stimulus complexity by

adding an extra piece of information to be processed by the
driver.

99

II) Left-turn and through signal illuminated section
configuration (COLORS). Several conditions were analyzed:
i) Concurring left-turn signal sections-~green arrow alone or
with a green ball-—simu1taneously illuminated with a through
signal red ball;

ii) Discording sections on the left-turn signal—-green arrow and
red ball-- simultaneously illuminated with a through signal
red ball;

iii) Concurring left-turn signal sections--same as i) above--
simultaneously illuminated with through signal green ball;
and finally,

iv) Special stimuli .

The definition of COLORS incorporates a combination of signal
display complexity traits, such as: concurrence or discordance of the
left-turn and through signal indications; concurrence or discordance of
simultaneously illuminated sections on the left-turn signal; and,
through signal color. Concurring left-turn and through signal
indications may enhance the meaning of a protected left-turn indication
(all illuminated signal sections on all signal faces are green) while
discording indications may make drivers hesitate while deciding about
the correct action (the driver needs to understand that although the
through signal shows a red indication he/she still has the right of
way). Simultaneous presence of a green ball and a green arrow on the
left-turn signal face might confuse drivers not sure whether they should
assume a permitted (green ball) or protected (green arrow) indication.

A red ball simultaneously illuminated with a green arrow on the left-
turn signal might confuse drivers perceiving two seemingly contradictory
messages.
Protected: minor errors-all drivers

Minor errors are associated with answers indicating subjects think
they have to yield the right of way while turning during the protected
phase.

Minor error rates are different across age groups (an = 5.595, Fm»
= .001) with driver comprehension declining with age. The 95% confidence
interval for mean young driver comprehension (24%) does not overlap with
that of drivers over 60 years of age (mean 51%). The mean for 30-45 year

olds is 37%, and that of 46-60 year olds is 39%.

the presence and message of supplemental signs was found to be

100

statistically significant (an = 22.17, FProb = .000), and the sign
"left-turn signal" performed better than the no sign condition (average
error rates of 31.1% and 37.8%, respectively). COLORS was found to be
statistically significant (an = 69.7, FPM = 0.000). Best performing
(i.e., lowest minor error rates) were displays with concurring left-turn
signal sections (green arrow alone, or in conjunction with green ball)
simultaneously displayed with a through red ball (average minor error
rate 23.5%). Next best were displays similar to the previous ones with
a through green ball (31.0%), followed by displays with discording
left-turn signal sections (green arrow and red ball), simultaneously
displayed with a through red ball (43.8%). Finally, special displays
(e.g., fast flashing green arrow) had the lowest comprehension rates
(53 . 4%) .

A model incorporating both age and supplemental sign showed
statistically significant results for both variables (an = 4.81, FProb =
0.003, and Fm,o = 22.16, FProb = 0.00, respectively) while the age by sign
interaction was not significant. Similarly, a model including age and
COLORS showed both to be significant (an = 5.55, FProb = 0.001, and an
= 70.50, FP,ob = 0.000 respectively) although there were no statistically
significant interaction results. Signal position (both horizontal and
vertical) and signal head arrangement were not found to significantly
affect minor error rates.

Protected: minor errors-older drivers

Both variables SIGN and COLORS were found to be significant (an =
5.82, FProb = 0.021, and Fnuo = 17.93 FProb = 0.000 respectively), and both
were found to affect all and older driver comprehension in a similar
manner. For example, locations with a sign "left-turn signal"
associated with lower minor errors for all drivers, were also associated
with lower minor error rates among older drivers as well.
Protected: summary and discussion

When all drivers were considered, AGE, COLORS, and SIGN were found
to be significant in explaining minor error variance. ANOVA examining

AGE simultaneously with COLORS or SIGN showed significant main effects,

101

but no interaction effects. COLORS classifies stimuli according to
illuminated lens configuration. Best comprehended are stimuli with
concurring left-turn signal sections but discording left-turn/ through
sections, followed by concurring left-turn sections with concurring
through sections. Discording left-turn sections with a through red ball
follow, with the special stimuli category being the least understood.
Thus, concurrence of illuminated left-turn signal sections is most
closely associated with better comprehension, while discordance of
left-turn with through signal indication also results in better than
average comprehension. Low comprehension associated with special
displays can be interpreted as being the result of i) unfamiliar stimuli
(fast flashing green arrow for example), or, ii) a confusing
supplemental sign ("left-turn must yield on green ball" when green ball
and green arrow sections are illuminated simultaneously). The presence
of the sign "left-turn signal" was found to be associated with lower
error rates.

COLORS and SIGN are also statistically significant when older
drivers are analyzed separately. Stimulus categories defined in COLORS
and SIGN affect older driver comprehension in the same direction as they
do all drivers, but their effects are more pronounced for older drivers.

Minor error rates increase with increasing age. Youngest and
oldest driver comprehensions are about 13 percentage points in opposite
directions from the grand mean (equal absolute differences), while
drivers 31 to 45 and 46 to 60 years old are about one percentage point
from the grand mean.

In summary, minor errors in the comprehension of protected
left-turn displays are primarily related to the concurrence or
discordance of illuminated left-turn signal sections among themselves;
the concurrence or discordance of illuminated left-turn and through
signal sections; and, finally, the presence of unusual left-turn signal
stimuli and the presence or absence of the supplemental sign "left-turn

signal."

102

Red stimuli comprehension

Nine stimuli for the red interval phase are included in the
database. All stimuli consist of a through red ball and either a red
ball or a red arrow on the left-turn signal. Since only correct and
serious error answers are possible for red interval stimuli (any
misinterpretation of a red phase stimulus will lead to the violation of
another driver's right of way--a serious error), the analysis is limited
to the discussion of serious errors.

A number of variables that may affect driver comprehension of a
red left-turn display, such as horizontal signal positioning (HPOSIT),
number of through signals (NoTHRU), and the presence of a supplemental
sign (SIGN), for which adequate information was available, were examined
individually and simultaneously with driver age (AGE).

Red: serious errors-all drivers

Statistically significant differences in serious error rates exist
among driver age groups (an = 4.59, an = 0.004). In general, driver
comprehension deteriorates with advancing age, with drivers 16 to 30
years old having a mean serious error rate of 1.4%, increasing to 8.3%
for drivers over 60 years old.

Left-turn signal horizontal position was found to affect driver
comprehension significantly (an = 6.44, FProb = 0.012). Signals placed
in the straight-ahead position performed worse (3.1%) than those placed
between lanes (0.4%). It is noteworthy that signal horizontal position
affects driver comprehension in a manner opposite to that identified for
permitted displays, where a placement in the straight-ahead position was
less likely to be misinterpreted. This might be due to a carry-over
effect from permitted displays: drivers may think that, since a green
ball has a different meaning for a driver turning left, the red ball
meaning is also different for the left-turning driver. Correct
interpretation of left-turn green balls depends on successful
differentiation from through green balls; by contrast, correct
interpretation of left-turn red balls depends on not differentiating

their message from through red balls. Thus, green balls are better

103

comprehended when placed further from the through signal and red balls
are better comprehended when placed closer to the through signal.

Driver age-horizontal position interactions were not found to be
statistically significant. No significant effects were identified for
number of through signals (NoTHRU) or the presence of the left-turn
supplemental sign (SIGN).

Red: serious errors-older drivers

Left-turn signal horizontal position (HPOSIT) was the only
variable that was found to significantly affect older driver
comprehension (an = 4.96, Fhob = 0.031). The straight-ahead position
is associated with lower comprehension scores (12.7%) than a placement
between lane lines (6.1%), a pattern consistent with that found for the
entire subject sample. Number of through signals (NoTHRU) and presence
of left-turn supplemental sign (SIGN) were not found to be statistically
significant.

Red: summary and discussion

AGE was found to have a statistically significant effect on serious
error rates. Driver comprehension deteriorates with age with the
youngest age group showing the best comprehension and drivers older than
60 years of age having the worst.

Comprehension of red interval stimuli is of paramount importance,
since any misinterpretations of its meaning may have grave consequences
while driver age differences were significant,
misinterpretations were relatively low. This part of the analysis
verified comprehension differences among age groups, and identified a
small but statistically significant advantage of a signal placement
between lanes over a placement in the straight-ahead position. A
possible explanation for the higher rate of misinterpretations of the
straight-ahead placement is that drivers may feel that the meaning of a
red ball on the left-turn signal differs from that on a through signal
so they can turn left whenever they have an opportunity to do so if a
signal is placed in a manner that will make clear that it addresses

left-turning drivers exclusively. Number of through signals and

104

presence of the sign "left-turn signal" were not found to have a
significant impact on driver comprehension.
Change stimuli

Nineteen change interval stimuli are included in the database.

Their common trait is a yellow indication on the left-turn signal
(either a ball or an arrow). A second lens (red or green ball) may also
be illuminated on the left-turn signal for some stimuli, and each of the
three through signal sections (red, yellow, or green ball) is
represented in the database.

In addition to age relations to serious error rates, THRUCOL
(through signal indication) and CHANGE (pertaining to illuminated
left-turn signal section configuration) were examined. For CHANGE there
were three conditions:

1) Only a yellow ball or yellow arrow section is illuminated;
ii) A yellow arrow is simultaneously illuminated with a red
ball; and,
iii) A yellow arrow is simultaneously illuminated with a green
ball.

The first two conditions are identical (transition from permitted
or protected phases respectively) and require the driver to stop and
wait for the signal to change. The last condition indicates a
transition from protected to permitted, i.e., the driver may continue
without stopping if there is no opposing traffic.

Change: serious errors-all drivers

In this instance driver comprehension was not significantly
different across age groups (me = 0.457, FProb = 0.712). It is
interesting, however, to note the irregular serious error pattern across
age groups for change interval stimuli: the youngest and oldest driver
age groups have higher serious error rates (average 2.8%) than drivers
between ages 31 and 60 (average 1.7%). Change interval stimuli, unlike
green and red ball indications that may be interpreted differently by
turning drivers, convey the unique message that a change in right-of-way
priorities is about to happen and that drivers, have to come to a stop,

regardless of which maneuver they are executing. Not enough information

was available to evaluate the significance of other variables in terms

105

of serious error rates.
Change: serious errors-older drivers

Not enough data was available to evaluate comprehension effects of

the presence of database variables.
Change: summary and interpretation-serious errors

Driver age was not found to have a statistically significant
relation with serious error rate. There was not enough information in
the database for a detailed analysis involving additional variables.
Furthermore, the relatively low serious error rates associated with
change interval stimuli make it clear that further investigation of
serious errors is not required, since any possible signal display
enhancements will have minimal benefits in terms of reducing serious
intersection accidents.
Change: correct answers-all drivers

Correct answer rates were found to be significantly different across
age groups (an = 6.138, FProb = 0.001). Drivers 16 to 30 years old have
the best correct answer rate at 80% while drivers over age 60 have the
lowest rate at 63%.

CHANGE was found to be statistically significant (Pam = 77.89,
FProb = 0.000). Best comprehended was a yellow ball or arrow (88%),
followed by simultaneous yellow arrow and red ball (80%), with
simultaneous yellow arrow and green ball in last place (52%). THRUCOL
was also found to be statistically significant (me = 43.72, FProb =
0.000). Sufficient data were available only for the red and green
through signal indications. A through signal red indication is better
comprehended (82%) than the green ball (67%).

An ANOVA including both age and CHANGE showed significant effects
for both age (an = 4.17, FProb = 0.007) and CHANGE (an = 76.77, Few =
0.000), but none for their interaction. A similar model including age
and through signal color had significant main effects (Pam = 6.82, FProb

0.000) as well as significant age/color

= 0.000, and rum = 43.05, Fa»

interactions (an = 3.02, FProb 0.031).

106

Change: correct answers-older drivers

When only older drivers were considered, CHANGE was found to be
statistically significant (an = 5.36, FProb = 0.007). The order of best
comprehended displays for older drivers was identical to that for all
drivers: best comprehended were left-turn signals with an illuminated
yellow ball or arrow (76%), while least comprehended were those with
simultaneously illuminated yellow arrow and green ball (52%). A yellow
arrow simultaneously illuminated with a red ball on the left-turn signal
was correctly understood by 65% of older drivers. Through signal
indication (THRUCOL) was not statistically significant (an = 0.34, FProb
= 0.562). Change interval stimuli incorporating a red through signal
indication are somewhat better comprehended (65%) than those with a
green through signal indication (62%).
Change: summary and interpretation-correct answers

Both through signal message (THRUCOL) and CHANGE are statistically
significant in explaining correct answer rates for all drivers, however,
only CHANGE is significant for older drivers. For the left-turn signal,
a yellow ball or arrow is best comprehended followed by the yellow arrow
alone or simultaneously illuminated with a red ball, while a yellow
arrow simultaneously illuminated with a green ball is least likely to be
comprehended. Left-turn change interval is best comprehended when
accompanied by through signal red ball. Left-turn change intervals are
less well comprehended when accompanied by through signal green ball
both among all and older drivers. Comprehension deteriorates with age,
and correct answer rate ranks are identical for levels of left-turn and
through signal indication variables among all and older drivers.
Change: summary and discussion

Serious error rates for change interval stimuli are at very low

levels (range 1.4%-2.4%), thus the differences among age groups have
little practical significance. Consequently, the rather low correct
answer rates (range 62.9%-80.3%) are due to a preponderance of minor
rather than serious errors. This observation suggests that change

interval stimuli may not be at the top of the practitioner’s agenda of

107

stimuli in need of improvement.

Although serious error rate changes with age are somewhat erratic,
correct answer rates consistently deteriorate with advancing subject
age. The only significant variable identified in the analysis is CHANGE
and the results are consistent with the complexity hypothesis in that
simpler displays (a single yellow ball or illuminated arrow section) are
better comprehended than displays with multiple illuminated sections
(yellow arrow and either red or green ball). Furthermore, among more
complex displays, concurrent indications (yellow arrow and red ball--
both indicating the transition to a stop) are better comprehended than
discordant displays (yellow arrow and green ball--one indicating a
transition to a stop, the other that the driver has the right of way)
requiring more rigorous interpretation. Display complexity hypotheses
are consistent with results for through signal indication: concurrent
displays (through red ball) are associated with better comprehension
than discordant displays (through green ball). These differences are
statistically significant for all drivers but not significant among
older drivers.

In summary, a practical method to improve correct answer rates for
change interval displays suggested by the analysis would be the use of
simpler displays, i.e., displays with a single illuminated yellow lens
(either ball or arrow). However, change interval displays do not appear
to pose a serious accident involvement threat to drivers due to
incorrect stimulus interpretation.

Flashing stimuli

Twenty-two flashing operations stimuli are analyzed in this section.
Since all three answer correctness levels are possible for flashing
operations stimuli, both serious error and correct answer rate analyses
are presented.

FLASH was defined in order to classify stimuli into three conditions
according to required driver action in response to particular flashing
operations configurations:

i) Stop. Proceed when there is no traffic in both directions

108

(in response to flashing red balls both on the left-turn and
through signals).

ii) Stop. Proceed if there is no opposing traffic (in response
to a flashing left-turn red ball and flashing through yellow
ball).

iii) Permitted left-turn (proceed with the left-turn without

stopping, unless you have to yield to opposing traffic, in
response to flashing yellow balls on both the left-turn and
through signals).

Through signal color (THRUCOL) and the left-turn signal indication
(LFTFLSH) were also examined. LFTFLSH has three conditions: a dark
left-turn signal face, a flashing yellow ball, and a flashing red ball.
Flashing: serious errors-all drivers

Although a serious error always means the violation of another
driver's right of way, driver actions counted as serious errors vary
across the three conditions of FLASH above. For example, not stopping
in i) is a serious error, while it is not necessarily for iii).
However, simultaneous examination of the three distinct messages in
flashing operations is useful in drawing conclusions about their
relative comprehension among drivers.

Although drivers younger than age 30 have lower serious error rates
(mean 10%) than those over age 60 (mean 15%), no statistically
significant differences were found between age groups.

There exist statistically significant comprehension differences
between levels of FLASH (Fflm = 83.74, FProb = 0.000). Stimuli requiring
the driver to stop and check both opposing and cross-street traffic
before proceeding show the worst performance (mean 27.2%), those
requiring a stop and check for opposing traffic before proceeding follow
(mean 7.6%), while those used to indicate permitted left-turns show the
best performance (mean 1.0%).

Through signal indication (THRUCOL) shows statistically
significant differences (an = 109.7, FProb = 0.000) with flashing red
ball showing much worse performance (mean 27.2%) than flashing yellow
ball (19%). Left-turn signal section type (LFTFLSH) is also

statistically significant (Frauo = 87.65, FProb = 0.000) with flashing

109

yellow ball best comprehended (0.4% errors) followed by dark left-turn
signal heads (12.4%), and flashing red ball worst (20.6%).

Three ANOVA models including variable AGE and one of the variables
FLASH, LFTFLSH, or THRUCOL showed no statistically significant age
effects and no significant interaction effects.

Flashing: serious errors-older drivers

Results based on older driver comprehension are similar to those
based on the entire sample: FLASH is statistically significant (an =
19.97, FPM = 0.000). Permitted stimuli are best comprehended (average
1.5%), stimuli conveying the message "stop for opposing traffic" follow
(11.4%) and, stimuli conveying the message "stop for both opposing and
cross traffic" were the least well comprehended (33.0%).

Through and left-turn signal indications are statistically
significant (an = 29.26, FProb = 0.000, and Fnuo = 36.07, FProb = 0.000
respectively). Comprehension differences follow the same patterns as
for all subjects with through flashing red ball less well understood
(32.2%) than flashing yellow ball (22.3%); flashing left-turn yellow
ball stimuli are best understood (0.0%), followed by a dark signal face
(11.7%) with flashing red ball stimuli least well comprehended (26.2%).
Flashing: summary and interpretation-serious errors

Driver comprehension of flashing mode stimuli does not differ
significantly across age groups. Stimulus message (FLASH) provides
insight into driver comprehension of flashing operations since it
provides information on the through and left-turn signal configuration.
By contrast, through and left-turn signal configuration (THRUCOL,
LFTFLSH) do not describe flashing operations sufficiently when examined
individually, since right-of—way rules depend on both variables, thus
findings on these variables do not have much practical use.

Based on FLASH, left-turn message stimuli were best comprehended,
followed by stimuli indicating that the driver has to stop and then
proceed if there is no opposing traffic. Least well comprehended were
stimuli indicating that the driver has to stop and then proceed if there

is no opposing and cross-street traffic.

110

Driver comprehension is significantly affected by THRUCOL and
LFTFLSH. Both among the entire sample and older drivers, a flashing
yellow ball on the through signal is better comprehended than a flashing
red ball. Flashing yellow ball stimuli are best comprehended by all and
older drivers, followed by dark signal sections, with the red ball
indication least well comprehended.

Driver comprehension deteriorates as signal message complexity
increases. The simplest display, a flashing yellow ball on the
left-turn signal is always concurring with the through signal indication
(can only be present when a flashing yellow is displayed on the through
signal) and is automatically associated with a measure of caution (the
left-turning driver knows he/she does not have the right-of—way), but
the driver only needs to check for opposing traffic and does not need to
stop first.

Next in complexity, the dark left-turn signal head directs driver
attention to the through signal display: there are no discording
indications that need to be interpreted, however, the driver may
mistakenly assume that left-turns have the right of way just like
through drivers do, since they receive information from the same signal
as through drivers.

Finally, the most complex stimulus, the flashing left-turn red
ball, means that the driver has to come to a full stop and must always
be considered in conjunction with the through signal: a through flashing
yellow indicates that the driver has the right-of—way over cross street
traffic, so he/she can concentrate on opposing traffic while a flashing
red indicates that both opposing through and cross-street traffic has
the right-of—way. A through flashing red indicates a sequence of
actions: i) stop, ii) check the through signal, and iii) decide which
intersection movement(s) have priority. On the other hand, a through
flashing yellow indication does not require actions i) and ii) and is
always associated with the presence of opposing traffic only (priority
over cross traffic is granted). As the signal message becomes more

complex, correct message interpretation becomes less likely, especially

111

for older drivers.

The superiority of dark left-turn signal face over flashing red
ball leads to the conclusion that eliminating flashing red left-turn
signal indications may enhance driver comprehension (with the added
benefit of energy savings).

Flashing: correct answers-all drivers

Correct answer rates were found to be statistically significantly
related to subject age (Fnuo = 8.56, FE“ = 0.000). Correct rates
starting at 55% for the youngest drivers decline to 50% for drivers 31
to 45 years of age, 38% for those between 46 and 60 years old and,
finally, 32% for drivers over age 60.

FLASH was also found to be statistically significant (an = 26.16,
FProb = 0.000) with the best comprehension (58% correct) associated with
stimuli flashing red balls on both the left-turn and though signals.
Permitted stimuli (only yellow ball sections flashing) and stimuli
requiring the driver to stop and proceed if there is no opposing traffic
had lower correct answer rates (34% and 37% respectively).

Left-turn signal indication is statistically significant (an =
15.14, Fm» = 0.000) with flashing yellow least well comprehended (37%),
dark left-turn signals following (41%), with best comprehended flashing
red indications (51%).

Through signal indication was found to be statistically
significant (an = 43.96, FProb = 0.000) with flashing red ball better
comprehended (58%) than flashing yellow ball (34%). Three models
including variable ACE and one of the variables FLASH, LFTFLSH, or
THRUCOL showed statistically significant main effects, but no
significant interaction effects.

Flashing: correct answers-older drivers

Stimulus message (FLASH) was found to be statistically significant
(Pam = 6.94, FProb = 0.002). Correct answer rates follow the same order
for older drivers as for all drivers i.e., best comprehended are stimuli
flashing red balls on both left-turn and through signals (43%), followed

by those flashing a red ball on the left-turn and a yellow ball on the

112

through signal (37%), and last, those flashing yellow balls only (17%).
Left-turn and through signal indications were both found to be
statistically significant for older drivers as well (an = 9.51, PP,ob =
0.000 and an = 8.46, FProb = 0.006 respectively). The left-turn signal
showing a flashing red ball is best understood (42%) followed by dark
signals (29%) and least well comprehended, flashing yellow ball (18%).
Through signal flashing red ball has a higher comprehension (45%) than
flashing yellow ball (23%).

Flashing: summary and interpretation-correct answers

Correct answer rates are influenced by stimulus message: best
comprehended are stimuli used to convey the message "Stop. Proceed if
there is no opposing and cross street traffic," followed by those
conveying the message "Stop. Proceed if there is no opposing traffic."
Least often correctly answered were stimuli conveying the message of a
permitted left-turn.

While this order is identical for all and older drivers, some
differences exist in correct answer rates between the two groups:
stimuli indicating "Stop. Proceed if there is no opposing traffic" have
virtually identical correct answer rates, but comprehension of the other
two message categories drops by approximately 14 percentage points for
older drivers.

Driver comprehension is also influenced by driver age with a
noticeable decline in correct answer rates with advancing age. Age
interactions with stimulus meaning, through, and left-turn signal color
are non-significant. It is noteworthy that dark left-turn signal faces
are associated with a lower correct answer rate than flashing red
displays, a result contradicting serious error rate findings.

Flashing: summary and discussion

Flashing operations stimuli comprehension is the poorest among
stimuli categories. The peculiarity of flashing indications lies in the
fact that, while correct answer and serious error rates are both
affected by stimulus message, they are affected in opposite directions:

flashing red balls on the left-turn and through signals, have the

113

highest serious error and correct answer rates, while permitted stimuli
have both the lowest serious error and correct answer rates. This
phenomenon is due to differing minor error rates among stimulus message
categories. Driver comprehension of stimuli conveying the message "Stop.
Proceed if there is no opposing traffic" is more similar to that of
permitted stimuli in terms of serious error rates (low) and closer to
that of flashing red balls only, in terms of correct answer rates
(high). Older drivers tend to stop when facing a flashing indication,
regardless of its color, thus their response is incorrect for a
permitted display (lowest correct rate), but initially correct for a
flashing red ball since they must come to a stop for such indications
(higher correct rates). Although older drivers grant the right of way
to opposing traffic (low permitted serious errors), they fail to
recognize the additional right of way obligation to grant the right of
way to cross street traffic when both left and trough flashing red balls
are present (highest serious errors).

Although stimuli presenting comprehension problems for older drivers
have been identified, no design variables (such as signal position and
number of through signals for example) have been identified in the
database that would help alleviate these problems. The use of dark
left-turn signal faces during flashing operations can be recommended if
the goal is to reduce serious error rates. It should be kept in mind
though that a decrease in correct answer rates (i.e., increase in minor
error rate )is to be expected. An increase in minor error rates may be
acceptable however, since it may lead to less severe accidents.
lntra-interval summary

Stimulus comprehension measured in correct answer rates has been
shown to be significantly related to driver age in every stimulus
category for which correct answer rates were examined. However, older
drivers do not differ from other age groups in terms of serious error
rates for all stimuli categories examined with the exception of red
stimuli. Notwithstanding that misinterpretations of red stimuli may

have dire consequences, the finding is somewhat moderated given that red

114

stimuli serious error rates are quite low for all age groups. Although
different stimuli characteristics were found to affect driver
comprehension depending on stimulus meaning and whether correct answer
or error rates are analyzed, a general conclusion is that simpler
stimuli are better understood by drivers. A summary of the most

important findings in chapter 4 is presented in table 4.1.

115

 

Table 4.1. Analysis of left—turn signal comprehension-summary

 

Subject characteristics analysis-significant variables:

 

Miles driven per year:

Older drivers drive less miles per year than any other age
group.
Female drivers drive less than their male counterparts.

Answer correctness relations with driver characteristics:

Drivers confident about their answers are more likely to
comprehend stimuli correctly.

Drivers that drive 15,000 miles per year are more likely to
comprehend stimuli correctly than those driving fewer or more
miles per year.

Older driver answers are more likely to indicate a serious
comprehension error.

Older driver answers are less likely to be correct.

Drivers of both sexes are equally likely to give answers
indicating a serious error.

 

Inter-interval analysis-significant variables:

 

 

AGE:

subject age

Comprehension deteriorates with advancing driver age in terms
of errors (higher for older drivers) and correct answers (lower
for older drivers), for every stimulus meaning.

Correct answer rate differences among driver age groups are
more likely to be significant than differences in serious error
rates.

MEANING: permitted, protected, red, change, flashing stimuli

Change interval stimuli are best comprehended (least errors),
followed by red, permitted, and flashing interval stimuli which
are the least well comprehended (most errors).

Red phase stimuli are best comprehended (most correct answers),
followed by change, protected, permitted, and flashing interval
stimuli which are the least well comprehended (least correct
answers).

Similar comprehension patterns were observed for all and older
drivers, with older drivers showing lower comprehension than
other age groups in terms of correct answer rates (low) and
serious error rates (high).

 

 

116

 

Table 4.1. (cont'd).

 

Intra-interval analysis-significant variables:

 

Permitted stimuli:

 

AGE: subject age

0 No comprehension differences in terms of error rates were found
among driver age groups.
- Comprehension deteriorates with advancing driver age in terms

of correct answer rates (lower for older drivers).
HPOSITION: left-turn signal horizontal position

0 Placement on the left—turn lane centerline is better
comprehended (lower error rates) than placement between left-
turn and through lanes (higher error rates) when entire subject
sample is analyzed. '

0 Horizontal left—turn signal placement was not found to affect
older driver comprehension in terms of serious error or correct
answer rates.

 

Protected stimuli:

 

AGE: subject age

0 Comprehension deteriorates with advancing driver age in terms
of minor error rates (higher for older drivers). Minor error
rate differences among driver age groups are significant.

SIGN: presence of sign "left turn signal"

0 Sign presence improves driver comprehension in terms of minor
errors (lower where sign is present).

COLORS: illuminated lens configuration

0 Concurring left-turn signal lenses are associated with lower
minor error rates than discording.
o Concurring left—turn and through signal lenses are associated

with higher minor error rates than discording.

 

Red stimuli:

 

 

AGE: subject age

0 Comprehension deteriorates with advancing driver age in terms
of serious errors (higher for older drivers). Serious error
differences among driver age groups are significant.

HPOSITION: left-turn signal horizontal position

0 Placement between left-turn and through lanes is better
comprehended (lower serious error rates) than placement in the
straight-ahead position (higher serious error rates).

 

 

117

 

Table 4.1. (cont'd).

 

Intra-interval analysis-significant variables:

 

Change stimuli:

 

AGE: subject age

0 Comprehension deteriorates with advancing driver age in terms
of correct answer rates (lower for old drivers).

CHANGE: left-turn signal configuration

0 Yellow ball or yellow arrow are best comprehended in terms of
correct answer rates (highest rates).

0 Yellow arrow simultaneously displayed with red ball is less
well comprehended in terms of correct answer rates.

0 Yellow arrow simultaneously displayed with green ball is the

least well comprehended in terms of correct answer rates
(lowest rates).

THRUCOL: Through signal color

0 Red ball stimuli are best comprehended in terms of correct
answer rates (highest rates).

0 Green ball stimuli are not comprehended as well in terms of
correct answer rates.

0 Through signal indication has no effect on older driver correct

answer rates.

 

Flashing stimuli:

 

 

AGE: subject age

0 No comprehension differences in terms of minor and serious
error rates were found among driver age groups.
0 Comprehension deteriorates with advancing driver age in terms

of correct answer rates (lower for older drivers). Correct
answer rate differences among driver age groups are
significant.

FLASH: required driver action for flashing ball indications

o Stimuli with a yellow ball flashing on both the left—turn and
through signal were associated with the lowest serious error
and lowest correct answer rates among flashing stimuli.

o Stimuli with a red ball flashing on both the left-turn and
through signal were associated with the highest serious error
and highest correct answer rates.

0 Stimuli with a red ball flashing on the left-turn signal and a
yellow ball flashing on the through signal were associated with
intermediate serious error and correct answer rates.

 

 

CHAPTER 5
FIELD DATA ANALYSIS

5.1. INTRODUCTION

Higher older driver accident involvement when turning left or
right at intersections has been identified by a number of investigators
as discussed in the literature review. The findings presented here are
based on the field data analysis, designed to parallel, to the extent
possible, the laboratory data investigation. Consistent with the
methodology described earlier, a general overview of field data is
presented first. The analysis focus becomes progressively narrower
examining relationships between through, left-turn, and right-turn
accidents and driver age, succeeded by comparisons between intersections
utilizing protected, permitted, and protected/permitted left-turn
phasing. A discussion on the relationships between driver age and left-
turn accident statistics with various approach geometric and signal
variables conclude the chapter.
5.2. MANEUVER ANALYSIS

Accidents during full color (referred to as "daytime" in what
follows) and flashing (referred to as "nighttime") signal operations
account for 86.6% and 13.3% of all intersection-related accidents,
respectively. Among daytime intersection accidents, 78% involve an at-
fault driver moving straight through the intersection, 4% a driver
turning right, and 17% a driver turning left (figure 5.1. and table
5.1).

The percent of accidents involving drivers moving through drops
with increasing age (figure 5.1), until it becomes 66% for drivers 60
years of age or older. The drop in through accident involvement is
accompanied by a simultaneous increase in left-turn and right-turn
accident involvement (from 17% to 28% and from 4% to 6% respectively) as

drivers age. Among drivers over 60 years of age, left-turn accidents

118

119

 

 

Percent of age group accidents

Figure 5.1. Daytime accidents

 

LEFT THROUGH RIGHT
Intersection maneuver

15-30 years old E 3045 years old 46-60 years old [BE 60+ years old

 

 

120

are 59% and right-turn accidents are 51% higher than average for

left-turn and right-turn accidents respectively.

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 5.1. Normal signal operation accident involvement
(percent)
LEFT THROUGH RIGHT TOTAL

All approaches

Average 17 78 4 100 '

Drivers 60+ 28 66 6 100
Permitted

Average 25 71 5 100 ’

Drivers 60+ 36 58 6 100
Permitted/protected

Average 16 81 3 100

Drivers 60+ 30 65 4 100 ‘
Protected

Average 9 86 4 100 °

Drivers 60+ 18 73 9 100
* Totals may not add up to 100% due to rounding errors

 

When the data are further stratified by left-turn phasing, similar
patterns are evident for approaches with permitted, protected, and
protected/permitted left-turn phasing, i.e., a heavy concentration of
through accidents declining with driver age and a simultaneous increase
in turning accidents with a preponderance of left-turn accidents.
However, each left-turn phasing type is associated with unique
characteristics, as summarized in table 5.1 and figures 5.2, 5.3 and
5.4. Permitted phasing is associated with the highest percentage of
left-turn accidents (36%) for drivers 60 years old or older while
protected phasing establishes the other (low) end of the spectrum with
left-turn accidents accounting for 18% of the accidents.
Protected/permitted phasing falls between the two extremes (30%).

These findings are consistent with those identified in the
literature review and are compatible with driver comprehension findings.
Protected displays are the simplest and drivers are more likely to

understand them correctly. Since drivers have the right-of—way during

121

 

 

Figure 5.2. Permitted locations

 

 

 

 

 

 

 

 

H

\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ \m
_________\__________________________________________________________________________________\___.,a m

 

    

lntersecTtionG maneuver

 

 

 

$523.... 956 one .o Eoocoa

old E 30-45 years old 46-60 years old E 60+ years old

15-30 years

 

 

122

 

 

Percent of age group accidents

Figure 5.3. Protected locations

 

80-i
701
60-1
504
40‘

301

 

111111

111:
IIILLILILII]
I'IIITIIIIJI

 

\V

 

1

11 11 A I
LL11 II III III] IIIILIIII LLI

1

 

11

IIIIIITI TIT IIUIIIIILH

TVIIIIII HI

1111

 

 

TI IITI

L11

 

 

 

1111 II III [III

1114
III_I II

 

 

 

 

W

LEFT THROUGH 1 RIGHT
Intersection maneuver

 

15-30 years old a 30-45 years old 46-60 years old m 60+ years old

 

 

 

123

 

 

 

 

RIGHT

 

 

 

3. ______________________________
HHHHHHHHHHH

 

 

 

 

 

OUGH
lntersecTtion maneuver

          

s\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\m

 

Figure 5.4. Permitted/protected locations

 

 

no mo mo « nw
% 8 7 6 w 4 3 2 1|

mEmEoom 935 one ..o «coated

15-30 years old a 30-45 years old 4660 years old m 60+ years old

 

 

124

the protected interval, it is highly unlikely that they will be at
fault while turning left. Permitted displays have been shown in the
comprehension analysis to be associated with a significant number of
errors among drivers (especially older ones): drivers mistakenly
thinking they have the right of way during the green ball interval are
likely to be the driver at fault in a left-turn accident. Permitted and
protected phasing (shown to be associated with higher and lower
percentage of left-turn accidents respectively) are present in
protected/permitted displays. However, permitted interval durations
usually dominate left-turn green time at such locations, providing a
possible explanation of left—turn accident percentages closer to those
of approaches with permitted rather than protected phasing.

Nighttime left, through, and right turn accident characteristics are
within a couple percentage points of those of daytime accidents (figure
5.5). Right-turn accident involvement remains approximately constant
with age (table 5.2 and figure 5.5), and, thus changes in the types of
accidents associated with different driver age groups are confined to
shifts between through and left-turn accidents. Patterns are similar to
those for daytime conditions with a decrease in through accidents while
left-turn accidents increase with increasing driver age. However,
drivers older than 60 have a lower percentage of nighttime (23%) than
daytime (28%) left-turn accidents. This finding may be attributed to a
number of factors, among which are: i) the presence of lower traffic
volumes during flashing operations; and, ii) a conservative older driver
response to flashing operation stimuli identified in the laboratory
analysis. Older drivers have lower correct answer rates due to minor,
not serious errors, i.e., they tend to give away their right of way when
facing a flashing indication. By contrast, older drivers are more
likely to think they have the right of way when facing a left-turn green
ball during daytime. The above findings (tables 5.1. and 5.2. and

figures 5.2. through 5.5) are consistent with the hypothesis of higher

125

 

 

 

 

 

 

 

 

 

 

 

 

   

m
m a __________________ asses. H
m aaaaagaaae w
m. \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\_ m
m. .\\\\\\\\\ m

 

 

11 I Al I Jl I I i _
m 0 0 m 0 0 O o 0 0
8 7 5 4 3

mEoEoou 966 one no Eoocoa

intersectionG maneuver

15-30 years old E 30-45 years old 46-60 years old m 60+ years old

 

 

126

accident involvement in more complex situations for older drivers. A
shift from accidents involving straight-moving vehicles (simple
rnaneuver) to those involving drivers attempting a turn (more complex
rnaneuver) is evident with increasing driver age. Among turning
movements, the most significant increase with driver age is for

left—turn accidents, both in terms of percentages and absolute numbers.

 

Table 5.2. Nighttime signal operations accident
involvement (percent)

 

 

 

 

 

 

Left Through Right Total
Average 15 80 5 100
Driver 60+ 23 73 5 100 °

 

 

* Totals may not add up to 100% due to rounding errors

 

 

Involvement ratio graphs for all, through, right-turn and left-turn
daytime, and nighttime intersection-related accidents are depicted in
figures 5.6 through 5.9. The abundance of data available for this

analysis allowed for a large number of age cohorts (table 5.3).

 

 

 

Table 5.3. Age cohorts used for figures 5.6 through 5.9
Age cohort Represented by value
up to 17 15

18 through 22 20
23 through 27 25
28 through 32 30
33 through 37 35
38 through 42 40
43 through 47 45
48 through 52 50
53 through 57 55
58 through 62 60
63 through 67 65
68 through 72 70
73 through 77 75
78 through 98 80

 

 

 

 

Involvement ratios were calculated for each age cohort shown in
table 5.3. Continuous lines connecting 14 distinct points at which
Dl/D2 ratios were calculated and a horizontal line D1/D2 = 1 indicating

the point at which drivers are neither over- nor under-involved in

127

 

”Ema—00.4 1.030th I ”Ema—00¢. 13¢. IXI

was .525

 

 

 

 

8 on 8 cm 9. on om

_ mm _ mm _ mm R me. A mm _ mm _ mu mo
r a
1m. a
TN
Lalo.
In
Two
1.

 

 

 

mozmc EoEo>_o>:_ EoEoom oEz>GD .m.m 0.59“.

ones ZCI/ IO

 

 

128

 

mkzwn=00< FIGE I_I wHZwQOO/x ..3< IXI

was .625

 

 

 

 

om on cm on ov om om
mm mm mm me mm mm m F
_ _ G _ _ _ _ _ _ A _ _ _ m.O
T F
Tm. F
...» IN
.3
. .. de
ilIi
Tm
1mm
To
me

 

 

 

wozmt EoEo>_o>:_ EoEoom oEimo Hm 059“.

0991:! 30/ IO

 

 

129

 

mHZwQQO< Em.— l+l wkzw900< ._._< .IXI

com .025

cm on om om CV 00 ON
mm mm mm mu mm mm

 

 

 

 

 

 

 

moan. EoEo>_o>:_ EoEoom 0.:ng .m.m 9.3mm

OIIBH ZG/ I-CI

 

 

130

 

w5m0_00< NEE—1.0.2 i+l ”Ema—09g 4.2 IXI

mam 32.5
cm on om om ov om om

 

 

 

 

 

 

 

 

mosh. EoEo>_o>:_ EoEoom 0.53292 Wm 0.5mm

OIIEH ZCI/ lCI

 

 

131

accidents are presented to facilitate visual pattern recognition and do
not signify the computation of "continuous" involvement ratios across
ages.

A consistent "U-shaped" pattern of younger and older driver
accident overinvolvement and "middle aged" driver underinvolvement is
clearly identifiable for all types of accidents. Thus, involvement
lines cross the equilibrium line twice in a driver's life, once
downwards (i.e., from over— to under-involvement) and once upwards
(i.e., from under- to over-involvement) as the driver ages. However,
differences exist among accident types for the ages at which involvement
lines cross the equilibrium line, in the rate of change of the
involvement ratio, and extreme values of over- and under-involvement.

As mentioned in the methodology chapter, age limits used in figures 5.6
through 5.9 are defined as a matter of "convenience" rather than based
on definitive changes in driving behavior occurring at particular ages.
In the same context, figures 5.6 through 5.9 cannot be used to pinpoint
exact ages when drivers become over- or under-involved in accidents or
calculate exact rates of change of involvement with age since there is
significant variability with regard to driving ability within age
groups, especially among older drivers. The figures are meant to
illustrate general tendencies occurring over decades throughout a
driver's life. A line representing all daytime accidents is present in
every figure to facilitate comparisons among different types of
accidents across figures.

The line representing all daytime accidents in figures 5.6 through
5.9 is based on 86.6% of the accidents in the database and shows young
driver (age cohort 15--see table 5.3 for definitions) involvement of
1.77 (overinvolvement). Risk declines with increasing age, reaches
equilibrium between the latter half of the third and beginning of fourth
decade, and a minimum (.72) in the fifth decade of a driver’s life.

From that point, there is a monotonic increase in involvement, and,

132

equilibrium is reached again between the seventh and eighth decades.
Drivers over age 78 have a maximum involvement of 1.92. Daytime through
accidents (representing almost three-quarters of the database-figure 5.1
and table 5.1) closely follow total daytime trends (figure 5.6) with a
nearly identical young driver involvement (1.76) that drops with age but
at a lower rate than that for total accidents, crossing the equilibrium
line at about the end of the third decade. Through-accident involvement
eventually crosses the total accident involvement line between the
middle of the fourth and fifth decades and represents the lowest
involvement among examined accident categories for all drivers after the
fourth decade. An all-age minimum involvement (.68) is reached around
the middle of the fifth decade and from that point involvement increases
with increasing age; equilibrium is reached around the eighth decade.
Drivers over 78 years of age have an involvement ratio of 1.41.

Right-turn accident involvement (figure 5.7) is the highest among
examined accident type categories for the youngest age cohort (2.18). A
sharp decline with age, leads to equilibrium during the third decade and
a minimum (.65) between the fourth and fifth decades. Starting during
the sixth decade, involvement increases in a step-wise fashion reaching
equilibrium between the sixth and beginning of seventh decades. A sharp
increase is observed during the eighth decade. Involvement becomes 2.68
for drivers older than 78 years of age, a value significantly higher
than that for all and through accidents for the same age group.

The most dramatic fluctuations with driver age among examined
accident categories (figure 5.8) are evident when considering left-turn
accident involvement. Starting at 1.74 for the youngest drivers,
involvement drops dramatically with age, reaches equilibrium during the
second decade and a minimum (.70) during the third decade. Experiencing
some fluctuation during the third and throughout the fourth decade, a
steady increase is noted during the rest of the driver's life.

Equilibrium is reached again during the fifth decade. Drivers older

133

than 78 years of age have an involvement ratio of 4.34, the highest
among all accident categories examined.

Nighttime accident involvement closely follows daytime patterns
(figure 5.9). Involvement is the lowest among the youngest drivers
(1.62) among all accident categories examined; equilibrium is reached
during the third decade; a minimum (0.69) is reached between the fifth
and sixth decades; and involvement fluctuates around the equilibrium
during the last half of the sixth decade and increases rapidly during
and after the eighth decade (1.94).

Maneuver analysis: summary and discussion

The figures in the preceding section serve to illustrate that
intersection accident involvement characteristics change with driver
age. The most important findings are highlighted below and followed by
suggestions for explanatory mechanisms,based on the literature review
and laboratory comprehension analysis results.

There is a shift from through to left-turn and right-turn daytime
accidents with increasing driver age. For nighttime accidents a shift
is observed strictly between through (decrease) and left-turn accidents
(increase) with right-turn accidents remaining constant with advancing
driver age. Nighttime left-turn accident increase is less pronounced
than daytime, a finding that corresponds to a relatively high incidence
of serious errors for permitted interval stimuli and low serious error
rate for flashing left-turn red ball stimuli among older drivers.

The percentage of daytime left-turn accidents at approaches with
permitted left-turn control is higher than that at approaches with
protected/permitted and protected control. These findings correlate
well with stimulus comprehension findings showing that protected stimuli
are better comprehended than permitted stimuli. Protected/permitted
location percentages are between those of permitted and protected
locations, an expected result, since part of the green time is treated

as a protected and part as a permitted interval.

134

An overinvolvement of younger and older drivers and
underinvolvement of "mid-aged" drivers has been observed for all daytime
and nighttime accidents, regardless of intersection maneuver (i.e.,
left-turn, through, right-turn). The age at which younger drivers cross
the equilibrium line (i.e., the line D1/D2 = 1) and that at which older
driver accident overinvolvement begins is different for each
intersection maneuver, with the most pronounced differences shown for
the left-turn maneuver, for which overinvolvement begins during the
fifth decade (earlier than any other maneuver) and remains more
pronounced than that of any other maneuver throughout the remainder of a
driver's life.

Older driver involvement ratios vary significantly depending on
intersection maneuver. Through accidents are the least hazardous for
older drivers (1.41) followed by nighttime signal operations (1.94),
right-turn (2.68) and left-turn (most hazardous--4.34). Older driver
accident involvement is consistent with the hypothesis of higher
involvement for more complex intersection maneuvers: The through
movement representing the simplest task for a driver in the intersection
environment is associated with the lowest involvement. The next higher
step in intersection maneuver complexity is that of turning right,
involving monitoring the traffic signal, while simultaneously
maintaining a safe distance from the leading vehicle, and observing
pedestrian traffic. Increased complexity of the mental and physical
tasks associated with a right-turn maneuver is reflected in higher
involvement ratios for right-turn accidents compared to the simpler
through maneuver. Finally, the left-turn maneuver, the most complex
task among the ones examined here, involves (depending on the type of
left-turn phasing) locating the left-turn traffic signal and
supplemental signs, and correctly interpreting their message, while
simultaneously monitoring leading vehicles and pedestrian traffic on the

cross street, identifying and safely using acceptable gaps in the

 

 

135

opposing traffic, and correctly identifying the turning path. Task
complexity places a heavy mental load on the driver and accident
involvement reflects the relative inability of the older driver to
successfully negotiate left turns compared to all other intersection
maneuvers.

Nighttime involvement ratios closely follow daytime accident
involvement trends with minor deviations. This finding appears to be
counter-intuitive given the common perception of increased older driver
accident involvement due to physical limitations, according to which
older drivers would be expected to have a higher accident involvement
during nighttime. For example, since older drivers cannot see as well
as their younger counterparts under nighttime (low intensity
illumination) conditions, they run a higher risk of hitting poorly
illuminated objects or running off the road. The absence of older
driver increased involvement in nighttime accidents can be explained as
a result of interactions of a number of factors: i) only the most
physically fit older drivers drive at night--those that feel they have
physical limitations do not drive; ii) older drivers commit fewer
serious comprehension errors (see definition in comprehension analysis
chapter) when facing a flashing red signal than any full color display;
and iii) the presence of lower traffic volumes during nighttime allows
for selection of longer acceptable gaps and more opportunity for
collision-avoidance maneuvers by drivers about to collide.

In summary, two trends have been identified as drivers age: a
shift from straight through to turning accidents and a higher
involvement ratio for turning accidents, more pronounced for left than
right turns. A higher percentage of older drivers are involved in left-
turn accidents at locations with permitted left-turn phasing than
locations with protected left-turn phasing. All identified trends can

be related to maneuver and signal display complexity.

136

5.3. LEFT-TURN ANALYSIS

It was just shown that there is a significant shift towards
left—turn accidents with increasing driver age. Furthermore, the
proportion of left—turn among all intersection accidents was shown to
depend on left-turn phasing type. Differences in left-turn accident
percentages among left-turn phasing types warrant separate examination
of accidents at approaches controlled by permitted, protected, and
protected/permitted phasing. (It should be noted that the exact instant
that an accident occurred during a cycle is not known thus an exact
apportionment of accident contribution for each interval displayed on a
left-turn signal is not possible.) In the absence of precise temporal
information, the assumption is made that right-of—way rules associated
with a particular left turn control scheme have an overall effect on
accident experience. Furthermore, it is assumed that most left-turn
conflicts occur during the green and change left-turn intervals.

Given this context, left-turn accidents are segregated into three
groups according to left-turn control (i.e., permitted, protected, and
protected/permitted), and individual involvement ratios are compiled for
each group. Age groups defined in table 5.4 are used for involvement
ratio graphs (figures 5.10 through 5.12) and do allow direct comparisons

with age groups used in the comprehension analysis earlier. Figures

 

Table 5.4. Age cohorts used for figures 5.10 through 5.12

 

Age cohort

 

15-30 years old
31-35 years old
46-60 years old

61+ years old

 

 

 

5.10 through 5.12 are similar to the ones presented previously with some
minor changes: i) a line representing left-turn accidents at all
approaches is present to serve as a reference; and, ii) since fewer

observations are available for each figure (e.g., left-turn accidents at

137

 

$20550... amt—Emma ...nJLi. WZOP<OOJ 4.2 II

was 32.5
+ 5 comma mflom oflmr

 

 

md

rmd

 

 

L1“

 

 

 

ed

moss. EoEw>_o>:_ EwEoom ERIE 0:550 .ofim 0.59m

OHBH ZCI/I-G

 

 

138

 

wZOF<OO._ owkokamm Tm: wZO_._.<OO._ j< I

one Lozen—
+ 5 oflmv 91.0” no.3

 

 

I oo

 

Tvé

 

 

 

 

NN

ooze. EoEo>_o>:_ EoEooe Sarto. oE=>eo . F Cm ocsnE

ONBH 30/ IO

 

 

139

 

m20:.<00.— EmmaPOmn. am.» wZO_._.<OO._ 4.2 I

one .635
+ E 8me mflom 00...:

 

 

 

 

IV.N

 

 

 

0N

ooze. EoEo>_o>c_ EoEooe Sarto. oEz>en .mfim o._:n_u_

ONES ZCI/ IO

 

 

140

permitted locations), fewer age groups were used than in figures 5.6
through 5.9 in order to include a significant number of data points in
each age group (some loss of detail is evident--compare lines
representing all left-turn accidents between figures 5.10 and 5.8 for
example).

Based on figures 5.10 through 5.12 older driver overinvolvement in
left-turn accidents is most pronounced at locations with
protected/permitted control (2.53), followed by locations with permitted
control (2.29). Locations with protected control are the safest for
that age group (1.87). Protected/permitted stimuli convey the most
complex message, requiring drivers to correctly understand right-of-way
rules associated with protected and permitted phasing and identify which
rules are applicable at any particular instant, thus confusion potential
is highest for such displays, leading to a higher older driver
involvement ratio.

Figure 5.13 is a comprehensive presentation of involvement ratios
intended to provide an overall view of older driver involvement in
intersection accidents. Information is stratified by: i) daytime or
nighttime accident occurrence; ii) approach left-turn phasing type
(i.e., protected, permitted, or protected/permitted); and, iii)
intersection maneuver. Although right turn and through accident
involvement ratios are presented for the sake of completeness, no
attempt is made to provide interpretations of results since these
maneuvers are beyond the scope of the present effort. Age group limits
used in figure 5.13 were set based on insights gained through the
overall field database analysis (figures 5.6 through 5.9) and the
following criteria: a) age groups should be broad enough to include
enough observations for the derivation of meaningful statistics; b) they
should be compatible with those used in the comprehension analysis for
cross-referencing purposes; and, finally, c) break points for age groups

should allow comparisons between under-involved "mid-aged" drivers

 

141.

 

no.0 no.0 «0.9 oo.F
+—o no-0e me-om op-op
co>aoces casu-ue we

 

 

 

 

 

 

 

 

 

~—.P h~.o no.0 pm..
oo.p mu.o so.o -.— +po oo-oe me-o~ op-op
+po oo-oe q-o~ o—-op
co>socee n on» mogoeocaae LL<
os_uugn_z

 

 

we." mo." en.o we."
+Po oo-oe mv-o~ o—-o—
co>zocee ccau-uwon

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

IL
se.. no.o ~o.o —~.p
+Po oo-oe me.om o—-op
Lo>3ocee coau-u; _¢
~—.F on.o no.o om.—
+—o oo-oe me.o~ o—-oF
oo.o n~.o co.o no.—
.3 8-3 .78 2 .2 _ 86.8%.. :<
co>aocee ; docs» oemu>eo
m..~ oo.p s~.o ~m.—
+90 oo.oe me-o~ o—.op
Lo>3ocee ccau.ueon
L

 

 

. ouucou
nueowmme swsuluuma use wo>smcee scauummwwucw an modwew ucme>Ho>cH .ma.m mwsnwm

 

 

1J42

 

 

 

 

 

 

 

up."
+po

magnu-u~o. uocasecoa ;u_: oco

n~.o
co-oc

nn.o c p
mvuow nplw

 

 

 

 

 

 

 

 

 

 

 

 

 

oo.p
+90

eccau.u~o. pouuouoc

no.0
oo.ow

50.9 no.—
e-o~ op.op
:u_x mco_ueoo4

 

 

 

 

 

 

 

 

 

 

 

 

 

nF.w
one

occau.uwo. pouu_5c

no.0
oo.oe

po.c Pm
e.o~ or
;u_: a:

 

 

 

 

oo.~ no.0 no.o op.—
+—o no-0q m¢.om 09-0—
cO>30¢es ccsu-uc mu
oo.o oo.o so.o Fo..
+90 oo-o¢ e-o~ op-o—
co>socee z 30cc»
nm.~ Fo.. m~.o o~.—
+Fo oe.o¢ no.0 o.-o.
co>socee ecsu-ueon
m~.. «o.o .n.o ~o..
+«o oo-oe me-o~ o—-e.
co>socee Ccsu-ucn_¢
oo.o eo.o co." «0.,
+—o 00.04 e-o~ o—-o.
co>socee z new
so.— oo.o Po.o mm.—
+Fo oo-ee me-o~ ow-o_
co>3ocee ecau-ueon
em.— on.o no.o m~..
+Fo oe.e< me-o~ op-o.
co>socee ccsu-ucn_¢
om.o m~.o so.o mo.—
+—o no-0e e.o~ op-o—
Lo>socee ; notch
o~.~ op.- e~.o mm.—
+po oo-oe me-o~ op-o.

co>soces ccsu-ueon

 

 

 

 

~—.P n~.o
+—o oo-oq

mo.o om.—
me-o~ n—-op

oozoeocaae ..<
ocomuecoao os_u>eo Leno»

 

 

 

 

.Ae.u:oov

.mH.m museum

 

 

143

(serving as a "control" group) and over-involved older drivers. Thus,
ages 46 and 60 used for the comprehension analysis were preserved as age
group limits since they fulfill all three of the above criteria.
However, it was felt necessary to establish a new age group limit at 20
years of age in order to separate overinvolved younger drivers from
those between 20 and 45 years of age who are clearly underinvolved. The
four established age groups (16 to 19, 20 to 45, 46 to 60, and 61+) are
used throughout the rest of this analysis.

Left-turn analysis: summary and discussion

Older driver left—turn accident involvement ratios are lowest
under protected, highest under protected/permitted, and intermediate
under permitted phasing. Drivers turning left under protected phasing
have a relatively simple task to accomplish: they need not worry about
opposing traffic interference, and can instead concentrate on keeping a
safe distance from the leading vehicle following the proper turning
path, and avoiding pedestrians crossing the cross-street. The task is
more complicated for drivers turning left under permitted phasing,
since, in addition to the previously detailed sub-tasks, they must also
identify and successfully use acceptable gaps in the opposing traffic.
Finally, protected/permitted left-turn phasing presents drivers with a
combination of the above listed situations. However, under this type of
left-turn control, the duration of the protected phase is usually a
small fraction of the permitted duration, and driver decisions become
even more complex than those involved in permitted phasing, since the
driver has to be aware of which left—turning rules are in effect (i.e.,
those pertaining to permitted or protected turns) at any given moment--a
process involving interpretation of signal information in addition to
monitoring other traffic behavior. Although accident involvement for all
and older drivers is affected in similar ways by left-turn signal
phasing, age differences become more pronounced as tasks of increasing

complexity are added to the left-turning maneuver. It is interesting to

144

observe that, although protected/permitted locations do not account for
the largest percentage of older driver left-turn accidents (figure 5.4),
they present the highest risk to older drivers. This finding points out
that, although protected/permitted phasing is harder to comprehend than
protected phasing for all age groups (thus the higher percent of left—
turn accidents), older drivers are much more adversely affected by the
complexity of the situation than other drivers (thus the highest
involvement ratios for protected/permitted locations).

In summary, older driver left-turn accident involvement ratios are
related to approach left-turn control. The simple protected control is
associated with the lowest involvement ratios, while protected/
permitted control, the most complex phasing scheme, is the highest.
Older drivers are underinvolved in through accidents and overinvolved in
right-turn accidents.

5.4. INVOLVEMENT RELATED TO SIGNAL, GEOMETRIC, AND ACCIDENT ATTRIBUTES

Given that older driver accident involvement appears to be
affected by left-turn control type, the analysis is now directed to
assessing the effects of geometric, signal, and accident variables under
each type of left-turn control.

Intersection accident involvement ratios are calculated for a
variety of variables for five accident categories: i) all daytime, ii)
all nighttime, iii) left-turn accidents at permitted locations, iv)
left-turn accidents at permitted/protected locations; and, v) left-turn
accidents at protected locations. A comprehensive accident involvement
table for all driver ages can be found in table 5.5. Accident
involvement ratios are presented for each of the previously defined
driver age groups, for a number of signal-, intersection geometry- and
accident-related variables. Where less than five observations are
present in the smallest involvement ratio numerator and/or denominator
for any age group, no involvement ratios are calculated and the line is
left blank. For example, not enough information was present for

actuated controllers at protected/permitted locations.

1J45

 

 

 

 

 

00.P 00.0 00.. 00 pouoouoca
00.~ 00.w 05.0 00.. 00.0 00.0 <0+00 0000.2coaxpouoouoca
00.~ 00.0 05.0 ~—.p 00J0 00.0 <0 .00 pouu_5coa\pouuouoca

<\z 00.~ 0~.- 05.0 00.— 00.? p0.0 00 + <0 pouoouoca
05.F 00.0 00.0 00.0 50.0 00.0 <0 pouoouoca

nn< 0P.P 00.0 p0.0 00 pouu_eLoa

coma-ssnmwcou Lacnma case-awed

00.~ 0p.p 05.0 00.0 00.0 ~5.0 0~.— 05.0 ~0.0 +5,

<\z 50.~ 50.0 F0.0 ——.0 00.9 ~5-0 <52 0«.— 00.0 00.0 0P-5
05J— 00.0 00.0 00.? 50.0 00.0 0 on a:

- 0.6000 comumcsu paquuoca

05.0 50.0 00.0 00.~ 05.. 05.0 0.." 00.0 00.0 +00

—0.~ 50.— 00.0 0~.~ 00.— 05.0 ~—.— 05.0 ~0.0 00-0~

<\z <52 50.~ 00.5 55.0 00.5 00.— 05.0 N..— 00.0 00.0 5--
00.9 00.0 00.0 0—.~ 00.5 55.0 00.F 00.0 00.0 -~ on a:

n-ooav sameness oeuummcoa

00.5 00.0 v0.0 . 00.5 00.— 00.0 "0.. 05.0 00.0 .uom 0~P.00

0—.~ 00.— ~0.0 00.~ 00.0 05.0 00.~ 00.5 05.0 0~.— ~5.0 00.0 .000 00

<\z 00.~ 0~.F 05.0 05.~ 00.5 55.0 05.. 50.0 00.0 .ooo 05-00
00.5 -.F 05.0 0~.~ 50.5 55.0 ~—.5 00.0 50.0 .uoo 00-00

59.2 Bud

00.~ 00.0 00.0 05.. -.— ~5.0 00.9 05.0 00.0 oueum o_.om

05.— ~5.0 «0.0 50.. 0—.— 00.0 00.0 05.0 00.0 pouesuo<

<\z 0 .P 00.0 00.0 50.~ 00.— 05.0 ~0.~ 05.. 05.0 0P." 00.0 ~0.0 .eo_cezoo:
to..ocucou

 

+—0 00-00 00-0N

+—0 00-00 00-0N

+—0 00-00 00-0~

+w0 00-00 00-0~

 

052m0_uu<
mz_55=0_z 44¢

mzo_5<000
005005080

 

020.5(000
zzwa\5oaa

 

mzo_~<uon
owh5_x¢wa

 

 

>420 052m0_uu< Cgaulumoa mt_5><0

 

mpzmo_uu<
m:_p><0 04¢

 

 

 

UOHUMH ucmem>do>flw UCUUHOW‘ .mom UHQWH

 

 

1J46

 

00.~ 00.—
5.~ 0F."

5—.— 05.0 No.0
00.5 55.0 00.0
p~.5 00.0 90.0

boo» 00-5-

uoow 00-0

«co» 0 on a:

comumeoa Leucowmcog 4ecnmm

 

00.. oo.o ~o.o

anemone
ucoooca uoz
once. ccsu-uwo. we coconuts

 

 

 

o~.~ .~.. ok.o ~0.. No.o oo.o 5.3-oco goes 0.3.030 so.-too.

00._ 05.0 50.0 ~0.p 00.0 ~0.0 00.~ 00.0 05.0 0~.~ 0—.— 05.0 NP.P 05.0 00.0. euoocom xez-ozu ozu noL-csoa
00.0 50.— 05.0 0... ~0.0 00.0 0o»

0000 comuooecouc_

0~.P 00.0 00.0 comuecwmm Laos-0~
05.5 00.0 05.0 <52 <\z <\z <\z not nc_;me.0 cuom
50.0 05.0 55.0 3o..o> nememe40 zoom
comuecsnmvcoo newsme00

00.~ 00.5 55.0 00.~ 00.0 05.0 55.. 05.0 00.0 004

<\z 50.. 00.0 00.0 -.— 50.. 00.0 <\z ~0.— 05.0 00.0 0004

oa>u omega oequuoca

 

+00 00-00 00-0N

+p0 00-00 00-0N

+—0 00-00 00-0~

+—0 00-00 00-0N

+00 00-00 00-0N

 

«pawn—uu<
mz_55:0~z 44<

mzo_5<uo4
0050050¢a

 

mzo_5<uo4
200055030

 

wzo~5<004
0055_x¢ma

 

 

>420 mhzw0000< Cusp-H504 wx~5><0

 

mpzw0_uu<
mz_h><o 44<

 

 

 

.Ae.udooc .m.m wands

 

 

1147

 

 

 

.o.o o~.° no.o mm.~ qm.. -.° oo.° m~.o so.o au_ .0 30:”
oo.. m~.o -.o no.~ ~n.c mo.o mo.~ no.o .~.a —~.~ n... o~.o oo.. o~.o «o.o “a:
-.. o~.c an.o n~._ ~0.o oo.o oq.~ 50.. 05.9 m_.~ n... m~.o o~.F .o.o No.o xga
, ou-figaa nae.

~e.. ”0.: «~.o ~o.~ «0.. o~.o on.c o~.o No.9 30cm
m~._ oo.o nh.o om.~ mo.o 05.0 om.m FP.. °~.o mo.o m~.° no.o c_a.
s... no.o mo.o no“

mF.. mm.o oc.o o~.. om.o ~n.c oq.~ oo.. m~.o c..~ mp.. o~.o w_.. om.o no.0 >u:o.o #0 .oo.u
50£u003

mo.~ .F._ .w.o oo.o mo.° oo.o o

Po.o oo.o oo.o ”5.. oo.o No.0 n¢.~ ao.° -.o «o.v oo.o mo.o u
oo.m mo.F Fo.o «m._ os.o .o.o a

”0.. om.o «m.o -.~ o~.F o~.o o~.P «m.o —o.o <

a0p-m udaauv odmwoua documm

 

+—0 00-00 m0-0N

+w0 00-00 m0-0N

+F0 00-00 m0-0N

+w0 00-00 m0-0~

+w0 00-00 m0-0N

 

mpzm0_uu<
mz_~p:0_z 44<

020—»<004
oupuw»0¢a

 

wzonh<004
zawmxh0¢a

 

wzo_~<000
cwhb~z¢mm

 

 

>420 mpzwo_00< Cusp-00; wz~h><0

 

mpzwo_uu<
mx_p><0 44<

 

 

 

.Au.ucoov .m.m ownma

 

 

1118

 

30LL< CON-.5 u

<0 44am coouo n 00

o4nou_.aa< «oz u «\z

>4maoocau4as_m vouac_s:44_ 44am conga uco zogg< coogo n me + <0
amazn vauuouoga no» Jagg< coogo .omozn vouu_e.oa gov veg mc_zmo4u u <0 .zu

.0.0 0L:m_0 o» Lowom .u4no_ga> m_zu Love: >gomuuou uco >4co u 44<

 

 

 

n~._ ~o.o mo.o ...~ no.o ~n.° oo.~ so.o n~.o ~«.~ ¢~.. ¢~.o o..F o~.o «o.o o..x
mo.P 00.0 .m.o 50.. oo.. on.o ~F.~ «o.. o~.o m_.~ mo._ -.o 0F.P o~.o No.o o4aeo.
, xow Lu>mun

om.o Nm.o oo.o om." .o.o ~n.o -.o .o.o oo.c amo_u co“ so..ou
~n.m n... oo.o oo.~ oo.. n~.o cgau .oaogas_

nn.. .~.o oo.o «o.. om.° no.. 00.? No.c «o.o um: can. goaoLae_
Nm.. mm.o «5.: oo.~ os.o mm.o no.~ o~._ ¢~.o om.~ .~.. ¢~.o mo.. oo.o mo.o xmx-wo-u;m_g u.a_> o“ ._.u
om.o ~¢.o Fo._ mc_uooam

comua4om> Lo>mga

oo.~ q~._ «o.c om.~ oo.o «~.o ~o.m 00.. ¢~.o Nq.~ «~.. n~.o se.. oF.. mo.o cgau-u*o. co-uao=
«¢.~ mo.o oe.o o~.F oo.o «o.o chau-u*o. aco-ga~¢

m~.° ~o.o ¢~.o oo.. mo.o no.o om.. oo.o _~.c ~m._ so.o .o.o on.. -.o cm.o cgau.o.mc<
.o.o oo.a .o.o mm.o o~.o oo.o uca.gau¢
mm.. om.o -.o mP.F mk.o no.0 u;m_.gum-o.mc<
25 «5393

 

+—0 00-00 m0-0N

+p0 00-00 m0-0N

+—0 00-00 m0-0N

+—0 00-00 m0-0~

+w0 00-00 m0-0~

 

mhzuo—00<
m24h>:0_2 44¢

020—><004
omhumhoxm

 

020~><004
z¢mA\>0¢m

 

020_><004
supp—tam;

 

 

>420 mhzwo_uu< ngu-u004 w24>><0

 

m>2m0_00<
mx_>><a 44<

 

 

 

..uyuquov .m.m manna

 

 

149

The goal of this analysis is to provide a comparison of older
driver accident involvement for different turn phasing strategies and
suggest the best among competing alternatives. The discussion in the
following paragraphs is focused on older driver involvement ratios
unless otherwise stated.

5.4.1. SIGNAL ATTRIBUTES

Older driver involvement ratios for left-turn accidents are
presented with respect to signal controller type, cycle length,
permitted phase duration, and protected phase duration. Possible
explanations for observed older driver involvement ratios are presented
within each section.

Controller type

Solid state controllers may alter phase sequence and/or duration
between cycles depending on intersection movement demands, while
actuated controllers may skip phases. By contrast, mechanical
controllers typically follow the same phase sequence with phase
duration, offset, and sequence changes occurring only at pre-determined
and specified times of the day. While mechanical controllers are
cheaper and of known reliability since they have been in service for a
number of decades, engineers justify the installation of more advanced
equipment based on, among other considerations, intersection delay
savings associated with more efficient traffic management. The question
arises, however, whether, by allowing alternate phasing sequences and
durations to occur frequently, older drivers who are used to a
particular phase pattern (sequence and duration) become confused and get
involved in relatively more accidents than at intersections equipped
with mechanical controllers.

It should be noted that the majority of mechanical controllers are
located at two-phase intersections and have typically been in place much
longer than actuated and solid state controller types. Thus, it is more

likely to find mechanical controllers at intersections where traffic

150

volumes have reached or exceeded capacity since the signal was first
installed. When saturated intersections are upgraded to more adequate
geometric designs, obsolete mechanical controllers are often replaced
with modern technology ones, therefore, it is quite possible, that there
are correlations between controller type, adequacy of geometric design
(e.g., presence of left-turn lanes) and other intersection parameters.
Due to database limitations, it was not possible to further stratify the
database and investigate the presence of such correlations.

The database analysis indicates that, for older drivers, actuated
controllers are associated with the lowest involvement for all daytime
(.99--table 5.6) as well as left-turn accidents under permitted (1.41)
phasing. Mechanical controllers are associated with the highest
involvement ratios (1.18 and 2.42 respectively) in both categories. No
significant differences exist among controller types for locations with
permitted/protected and protected phasing. Thus, it appears that
actuated controllers have beneficial rather than detrimental effects on
older drivers both in terms of all daytime and left-turn accidents at
permitted locations. It should be kept in mind though, that these

findings may be strongly related to more adequate intersection geometric

designs as discussed above.

 

 

 

 

 

 

 

 

 

 

Table 5.6. Older driver accident involvement ratios for different

siqnal controller types

LEFT-TURN ACCIDENTS
CONTROLLER ALL DAYTIME
TYPE ACCIDENTS
PERMITTED PROTECTED PROT/PER

Mechanical 1.18 2.42 1.74 2.57
Actuated 0.99 1.41 1.73 (1)
Solid State 1.09 1.73 (1) 2.50
(1) Insufficient data for analysis.

 

Cycle length

Cycle length is a surrogate for the number of change intervals and

stop—and-go procedures per hour.

A higher number of accidents

 

151

associated with a shorter cycle could be due to a larger number of
vehicular conflicts associated with the change interval and stop-and-go
operations. On the other hand, a higher number of accidents associated
with longer cycles may be due to more opportunity for vehicular conflict
during phases other than the change interval such as right turn on red
and permitted left-turns during which vehicular conflicts are more
likely to occur. The data indicates that older driver involvement
increases with increasing cycle length up to 80 seconds and drops
dramatically only for cycle lengths 90 seconds or longer. These findings
may indicate that there exists an equilibrium between the aforementioned
factors (i.e., number of change intervals and phase duration). Thus,
for shorter cycles, phase duration seems to be more influential
(accident involvement increases with phase duration); for longer cycles
number of change intervals becomes more important (less accidents for
fewer change intervals, despite longer permitted phase durations). The
critical (highest involvement) cycle length for older drivers is
different for different accident categories: a decline in involvement
ratios is evident for cycle lengths over 90 seconds for all accidents
and cycle lengths over 80 seconds for left-turn accidents at permitted

locations (table 5.7).

 

Table 5.7. Older driver accident involvement for different cycle

 

 

 

lengths.
LEFT-TURN ACCIDENTS
CYCLE LENGTH ALL DAYTIME
ACCIDENTS

PERMITTED PROTECTED PROT/PER
45-60 sec. 1.12 2.24 (1) 1.88
65-75 sec. 1.15 2.79 (1) 2.50
80 sec. 1.20 2.39 2.13 2.84
90-120 sec. 1.01 1.00 1.84 (1)

 

 

 

 

 

 

 

(1) Insufficient data for analysis.

 

Available information for older driver involvement in left-turn
accidents under protected/permitted and protected phasing seems to

confirm the suggested pattern.

152

Permitted duration

Left-turn accident involvement is directly related to permitted
interval duration, both for locations with permitted and locations with

permitted/protected phasing (see table 5.8) with an exception for

 

 

 

 

Table 5.8. Older driver accident involvement for different

permitted phase durations

LEFT-TURN ACCIDENTS

PERMITTED ALL DAYTIME
DURATION ACCIDENTS
( sec , ) PERMITTED PROTECTED PROT/PER
up to 21 1.30 2.15 1.95
22-27 1.12 1.86 2.37
28-35 1.12 2.29 N/A 2.81
36+ 1.19 2.99 3.19

 

 

 

 

 

 

 

permitted phase durations of 22-35 seconds that are associated with the
lowest involvement (1.12).

Older driver involvement ratios seem to be related to conflict
opportunity: the longer the permitted phase, the more opportunity exists
for conflicts with opposing traffic, and the higher involvement ratios
become.

Older driver left—turn accident involvement at locations controlled
by protected/permitted phasing is dominated by permitted phase conflicts
(i.e., accident involvement increases with permitted phase duration).

Protected duration

Older driver involvement increases with increasing protected
duration for all daytime accidents, and left-turn accidents under
permitted/protected phasing (table 5.9). However, for left-turn
accidents under protected phasing, involvement for a duration between 7
and 10 seconds (2.67) is higher than that for a duration longer than 11
seconds (2.50). Longer protected duration is an indication of a higher
demand for left-turns, a situation that may lead to left-turn lane (or
bay) overflow into an adjacent through lane depending on a number of

factors (such as presence of signal coordination, platoon arrival

153

pattern, leading or lagging left-turn phase), thus potentially creating
conflicts with through traffic. In addition, if right-turn-on-red is
allowed, there is a higher potential for conflicts between vehicles
turning left during the protected phase and opposing traffic turning
right on red. However, the simultaneous influence of other factors such
as protected left-turn phase type, left-turn signal lens configuration,
and concurrence/discordance with through signal indication need to be
addressed as well in order to arrive at a better comprehension of older
driver involvement relations to protected phase duration.

Unfortunately, not enough data is available in the present database to

allow further breakdown of the data.

 

 

 

 

 

 

 

 

 

 

Table 5.9. Older driver accident involvement for different

protected phase durations

LEFT-TURN ACCIDENTS

PROTECTED ALL DAYTIME
DURATION ACCIDENTS
( sec , ) PERMITTED PROTECTED PROT / PER
up to 6 1.00 (1) 1.78
7-10 1.16 N/A 2.67 3.11
11+ 1.24 2.50 3.33
(1) Insufficient data for analysis.

 

 

Signal attributes: summary and discussion

Actuated control was found to be associated with the lowest
oldest driver involvement and mechanical controllers with the highest
for all and left-turn daytime accidents. However, these findings do not
imply that a mere substitution of mechanical controllers with modern
technology ones will lead to reduced involvement ratios. This variable
provides a beginning point for a host of investigations aimed at
identifying the actual causative factors behind the observed statistics.
For example, actuated controllers are more likely to be present at
intersections with improved geometric design and mechanical controllers
are more likely to be present at intersections in need of design

improvements. Thus, observed statistics may reflect comparisons of

154

situations where left-turns are constrained due to high volume to
capacity ratios or inadequate design with situations where left-turns
face no such constraints. Similarly, signal timing at approaches with
mechanical controllers may be more likely to be obsolete (not reflect
current traffic conditions) than approaches with modern controllers.
However, an investigation of these factors is beyond the scope of this
study.

Accident involvement in all and left—turn daytime accidents at
permitted and protected/permitted locations initially increases with
increasing cycle length and then drops dramatically. Both phase
duration and number of change intervals per hour influence involvement
ratios. Initially, as cycle length increases, increased phase length
provides more opportunity for conflicts, but eventually, longer phases
that provide more and/or longer gaps in the opposing traffic and a
smaller number of change intervals during which conflicts may occur
dominate, driving involvement ratios downward.

For similar reasons, increasing permitted interval duration is
associated with increased daytime left-turn accident involvement ratios
at permitted and protected/permitted locations up to a certain duration,
after which accident involvement decreases.

In summary, the higher the opportunity for conflicts either due to
the duration and nature of a particular interval (e.g., permitted
interval) or the saturation of intersection movements (e.g., obsolete
geometric design—mechanical controllers), the higher older driver
involvement ratios become.

5.4.2. SIGNAL FACE AND PHASE TYPES

In view of the findings about the significance of signal display
complexity on driver comprehension, differences between three and four
stacked section displays operating under protected and protected/
permitted phasing are investigated. Effects of leading versus lagging
protected left-turn phasing on older driver left-turn accident

involvement are also analyzed. The section concludes with a discussion

155

of the effect of nighttime signal operation configuration on older
driver accident involvement.

Signal section configuration

Three alternate protected phasing signal face arrangements are
present in the database (table 5.10): protected green arrow (stacked
3), protected green arrow simultaneously illuminated with a green ball
(stacked 4), and protected green ball (in split phasing--a1so a stacked-

three configuration).

 

Table 5.10 Older driver accident involvement for different
protected phase signal face configurations

 

 

 

 

 

 

 

 

LEFT-TURN ACCIDENTS
PROTECTED ALL DAYTIME
SIGNAL ACCIDENTS
FACE PERMITTED PROTECTED PROT/PER
CONFIGURATION
PROTECTED GA“) 0.98 1.73
stacked 3
PROTECTED GB+GA 1.30 2.63
stacked 4 N/A 2.43
PROT/PER FR/GA 1.12
Michigan 2.68
display
PROT/PER GB+GA 1.39
stacked 4
Protected GB 1.53 (2)
(1) GA = Green arrow, GB = Green ball,
FR = Flashing red, / not simultaneously illuminated
+ = simultaneously illuminated

 

(2) Insufficient data for analysis.

 

 

Comparisons among all daytime accidents at protected locations
show that Older driver accident involvement is lowest (.98) for common
stacked-three displays with a protected green arrow (table 5.10)
followed by the stacked three Michigan display (flashing red ball for
the permitted and steady green arrow for the protected phase-—D1/D2 =
1.12). Stacked four (fourth level green arrow) displays are associated
with higher involvement ratios (1.30) than stacked—three protected green
arrow, but lower than protected green ball displays (green ball
displayed simultaneously with red ball for all other intersection

movements) which has the highest involvement ratio (1.53) among all

156

protected displays. Stacked-four displays are associated with higher
involvement ratios (1.39) than the Michigan display at permitted/
protected locations.

As shown in table 5.10 adequate data for left-turn accident
involvement is available for four displays, two used for protected/
permitted and two for protected phasing. Between the protected phasing
displays, the Michigan configuration performs better (2.43) than the
alternative stacked four display (2.68). The common stacked three
display performs better (1.73) than the stacked four display (2.63) for
protected locations.

Poor older driver performance associated with the protected green
ball display may be due to confusion about the unusual situation: the
driver has no way of knowing that opposing traffic faces a red phase
(i.e., no supplemental sign is provided, nor is he/she given an
indication of the change from the protected to a permitted interval,
since drivers face a green ball throughout the period Opposing traffic
faces either a red or a green ball). Among displays using arrow
sections, the simplest ones (stacked three) outperform the more complex
(fourth level green arrow) with two simultaneously illuminated lenses
(green ball and green arrow) in older driver total accident involvement.
Simpler (stacked three--common and Michigan) displays outperform the
more complex (more sections, two simultaneously illuminated sections)
stacked four displays in terms of left-turn accident involvement as
well.

Protected phase type

The choice between leading and lagging left-turn phasing is often
a matter of local policy or convenience, since no definitive accident
patterns have been established for the two phasing sequences. Table
5.11 is a summary of involvement ratios based on the field database.

Leading protected phasing is associated with a lower Older driver

involvement than lagging for all accident categories presented in table

157

5.11. Differences are small for all daytime accidents, somewhat higher

for left-turn accidents at protected locations, and particularly large

 

Table 5.11. Older driver accident involvement for different
protected phase types

 

 

 

 

 

LEFT-TURN ACCIDENTS
PROTECTED ALL DAYTIME
PHASE ACCIDENTS
TYPE PERMITTED PROTECTED PROT/PER
Lead 1.02 1.91 1.22
Lag 1.17 N/A 2.36 2.93

 

 

 

 

 

at protected/permitted locations where leading protected phasing is
associated with an involvement ratio of 1.22 compared to 2.93 for
lagging protected phasing. The higher left-turn accident involvement at
lagging protected phasing locations may be due to conflicts with
opposing through vehicles attempting to use the change interval

succeeding the permitted interval.

Flashing configuration

Field data analysis results presented in table 5.12 and indicate
that locations where both the through and left-turn signal flash yellow
balls have the lowest nighttime accident involvement ratios (.97).
Locations where both through and left-turn signals flash red balls have
higher ratios (1.14) and locations with 24-hour operations have the

highest involvement ratios (1.23).

 

Table 5.12. Older driver accident involvement for different
nighttime signal configurations

 

FLASHING CONFIGURATION ALL NIGHTTIME ACCIDENTS

 

 

 

 

Both flashing yellow 0.97
Both flashing red 1.14
24-Hour operation 1.23

 

These findings are in line with comprehension analysis findings of
a higher incidence of Older driver serious errors at locations with
flashing red than locations with flashing yellow balls.

It may be initially surprising to find that signals under twenty-

158

four hour full color operation are associated with the highest nighttime
accident involvement ratios for older drivers. However, it should be
kept in mind that twenty-four hour signal Operations are usually
reserved for higher nighttime volume intersections. Thus, older drivers
may be at a disadvantage at such locations due to shorter available gaps
when turning left (higher Opposing volumes), and glare from oncoming
vehicle headlights that contributes to difficulties with judging
oncoming vehicle speeds and distances and identifying proper turning
paths.

By contrast, flashing yellow indications on both the left-turn and
through signals are usually present at lower volume locations and thus
associated with fewer glare problems. Drivers are required to yield
only in the presence of oncoming traffic.

Flashing red indications however, require the most complex mental
processing and driver action sequence: the meaning of the left-turn
flashing red indication depends on the flashing through indication; for
a flashing red through signal drivers must come to a stop and proceed
only if there is no opposing through and cross-street traffic. Thus, at
such locations drivers have to check for traffic on three approaches
(compared to one needed for flashing yellow indications). In addition,
cross-street traffic speed judgement requires more physical effort
(i.e., turning the head in wider angles--a task demanding longer time
for older drivers-—and judging speeds of vehicles approaching in two
opposite directions) than opposing traffic speed judgment.

Higher maneuver complexity is reflected in higher involvement at
locations displaying flashing red compared to those displaying flashing
yellow indications. A possible explanation for the Observing the
highest older driver involvement in left-turn accidents at locations
with 24-hour signal operations may be found in higher nighttime traffic
volumes and the ensuing lower availability of gaps in the opposing
traffic commonly present at such locations. Older driver problems with

glare and judging opposing traffic speeds may be compounded to a higher

159

degree than those of other driver groups as nighttime traffic volumes
increase.

Signal face and phase types: summary and discussion

Simpler three-section protected displays are associated with lower
older driver left-turn and total daytime accident involvement ratios
than stacked four (fourth level green arrow) displays. Simpler displays
are associated with lower involvement ratios both when used for
protected and protected/permitted phasing. This result is compatible
with comprehension analysis findings of better comprehension scores
associated with simpler displays.

Older drivers have lower involvement in total and left-turn
accidents at locations with leading protected phasing than at locations
with lagging phasing. Higher lagging location involvement may be due to
conflicts with opposing through traffic attempting to use the clearance
interval between the permitted and protected phases.

During nighttime operations, locations where flashing yellow balls
are used for both the left-turn and through movements are associated
with the lowest older driver accident involvement, followed by locations
with flashing red balls for both left-turn and through movements. The
highest involvement was observed at locations with 24-hour full color
operations. Results for 24-hour operations locations may be due to
higher volumes at such locations as mentioned previously. Flashing
Operations involvement ratio results are compatible with comprehension
analysis results, i.e., a higher involvement occurs at approaches
controlled by displays with lower driver comprehension scores.

5.4.3. GEOMETRY AND LEFT-TURN SIGNAL POSITION

Differences in involvement ratios between approaches controlled by
mechanical and those controlled by solid state actuated controllers
identified in previous sections are believed to be associated with
modernization of geometric design and signal control parameters rather

than direct controller effects on accident experience. Thus, it is

160

desirable to identify relations between intersection geometry
information and involvement ratios. The present section is focused on
the analysis of the effect of various intersection types, presence of
left-turn lanes, left-turn signal horizontal position, and number and
position of signals (signal profile) on involvement ratios. Possible
effects of these variables on involvement ratios are discussed.

Intersection type

Older driver daytime accident involvement is lowest at Tee (1.10--
table 5.13) intersections and four-leg intersections of two-way streets
(1.12), and significantly higher at intersections of two-way with

one—way streets (1.62). A complete reversal of older driver involvement

 

Table 5.13. Older driver accident involvement for different
intersection types

 

 

 

LEFT-TURN ACCIDENTS
INTERSECTION ALL
TYPE DAYTIME
ACCIDENTS PERMITTED PROTECTED PROT/PER
Tee 1.10 (1) (1) 4.00
2-way/2-way 1.12 2.25 1.92 2.39
2-way/1-way 1.62 (1) (1) 2.20

 

 

 

 

 

(1) Insufficient data for analysis.

 

 

 

is recorded for left-turn accidents at protected/permitted locations
where involvement is highest at Tee intersections (4.00), lowest at
two—way with one-way street intersections (2.20), and intermediate at
intersections of two-way streets (2.39). Data for only one intersection
type are available for left-turn accidents at locations with permitted
and protected control, thus comparisons among intersection types are not
possible for these accident categories.

Additional information is necessary in order to interpret these
findings, especially the extremely high involvement ratio associated
with left-turns at Tee intersections which seems to be counter-
intuitive, since fewer vehicular conflicts exist at such intersections

and it seems reasonable to expect lower, not higher involvement ratios.

 

 

 

161

Differences between intersections of two-way streets and those of one-
way with two-way streets also need to be investigated further in order
to identify their causes.
Presence Of left-turn lanes

The presence of a separate left-turn lane is associated with lower

older driver involvement in all accident categories under examination

(table 5.14--not enough data were available for protected left-turns for

 

Table 5.14. Older driver accident involvement presence of left-
turn lane

 

 

 

 

LEFT-TURN ACCIDENTS
LEFT-TURN ALL DAYTIME
LANE ACCIDENTS
Typg PERMITTED PROTECTED PROT/PER
Not present 1.16 2.43 (1) 3.00
Present 1.09 2.22 1.84 2.49

 

 

 

 

 

 

(1) Insufficient data for analysis.

 

comparisons). Intersection movement separation with the provision of a
left-turn lane reduces traffic conflicts between through- and
left-turning vehicles resulting in fewer accidents both for left-turning
and through vehicles. Furthermore, a survey of older drivers identified
in the literature review (71) shows that older drivers are particularly
concerned with identifying the proper lane from which to turn. Thus,
provision of a left-turn lane marked as such helps eliminate this
problem, and older driver confusion decreases (since they can properly
position their vehicles for a turn), and, consequently, involvement

ratios become lower.

Signal horizontal position

A left-turn signal positioned zero to three feet to the left of the
left-turning driver is associated with the lowest older driver accident
involvement for all categories of left-turn accidents (table 5.15).
Involvement increases with increasing signal distance from the

left-turning driver.

 

 

 

162

 

Table 5.15. Older driver accident involvement for different left-
turn signal horizontal positions

 

 

 

LEFT-TURN ACCIDENTS
SIGNAL ALL DAYTIME
HORIZONTAL ACCIDENTS
pQSITION PERMITTED PROTECTED PROT/PER
Up to 3' 1.21 2.18 1.71 2.48
4’-16' 1.09 2.43 2.88 2.52
17’-38' 1.17 2.56 (1) (1)

 

 

 

 

 

 

(1) Insufficient data for analysis.

 

 

Signals positioned closer to the driver eliminate the need to scan
wide areas in order to locate the left-turn signal. Thus, signals can
be detected sooner and older drivers that have narrower peripheral
vision can monitor other intersection conditions pertinent to the left—
turning task easier. A lower scan time and the ability to easier
monitor intersection conditions lead to a better accident performance.

Signal profile

Signals addressed specifically to drivers turning left may
ameliorate poor driver performance by eliminating confusion regarding
right-of-way rules applicable to left turns. Signal heads addressed to
both through and drivers turning left may cause drivers to become
confused, especially during the permitted phase when the same indication
(green ball) on the same signal head must be interpreted differently by
drivers, depending on the intended maneuver.

The most common signal profiles found in the database are shown in
table 5.16. Profile A with three signal heads (one addressed to left—
turning drivers, one to through drivers, and one to both through and
drivers turning right) corresponds to approaches with three lanes.
Profile B corresponds to two-lane Tee and approaches intersecting with
one-way streets using two signal heads, profiles C and D to two-lane
approaches with a left-turn and shared through-right turn lanes.
Profile D approaches use two left-turn signals.

For left-turn accidents under permitted phasing, profile A

locations are associated with a higher Older driver involvement (2.22--

 

 

163

 

Table 5.16. Signal profile configurations

 

SIGNAL
HEADS SIGNAL ADDRESSED TO DRIVERS: PROFILE
PRESENT

 

Turning left
4 4 ft Driving through A
Driving through or turning right

 

q f Turning left ’ B
Driving through

 

4, ?> Turning left C
Driving through or turning right

 

41 1 it Turning left (two signals) D
Driving through or turning right

 

 

 

 

 

table 5.17) than profile D locations (2.03). Thus, locations with two
left-turn and one through signal outperform locations with one left-turn
and two through signals. However differences between the two signal
profiles are not dramatic. These results agree with comprehension
analysis findings that the number of through signals was not

significantly related to driver comprehension for permitted displays.

 

 

 

 

 

 

 

 

 

 

 

Table 5.17. Older driver accident involvement for different signal
profiles
LEFT-TURN ACCIDENTS
SIGNAL ALL DAYTIME
PROFILE ACCIDENTS
PERMITTED PROTECTED PROT / PER
A a. 9 f» 1.20 2.22 (1) (1)
B -q f 1.34 (1) (1) 3.60
c o, 1’ 1.04 (1) 1.78 2.43
D 4, 1 it 0.96 2.03 (1) (1)
(1) Insufficient data for analysis.

 

Older driver left-turn accident involvement at profile B locations
under protected/permitted control (3.60) is higher than profile C

locations (2.43). This result concurs with previous findings about

164

higher accident involvement at Tee intersections. However, it is not

clear why this phenomenon occurs. Not enough information is available
to allow comparisons of alternative signal configurations for Tee and

other intersection types.

In summary, configurations using a single through signal have lower
older driver left-turn accident involvement ratios at permitted
locations than configurations using two through signals at intersections
of two-way streets. Older driver left-turn accident involvement at Tee
intersections is significantly higher than that at intersections of
one-way and two-way streets when a single through signal is used. The
underlying cause Of this latter phenomenon remains unclear.

Geometry and left-turn signal position: summary and discussion

Involvement in left-turn accidents under protected/permitted
phasing is highest at Tee intersections, lowest at two-way with one-way
intersections and intermediate at intersections of two-way streets.

Left-turn signals positioned Closer to the driver are associated
with lower total and left-turn daytime accident involvement.
Involvement increases with left-turn signal distance from the driver.
These findings are attributed to less physically demanding and more
efficient visual scanning patterns associated with displays located
closer to the driver.

Left-turn accident involvement is higher at intersections of
two-way streets using two through signals than those using one through
signal under permitted phasing. This finding concurs with comprehension
analysis results. Better comprehension of stimuli using a single
through display was attributed to lower stimulus complexity. Lower
accident involvement, therefore, may be directly attributed to better
comprehension of simpler stimuli.

5.4.4. ACCIDENT AND DRIVER CHARACTERISTICS
The hypothesis that stimulus complexity is related to driver

comprehension of signal displays was shown to hold true on many

 

 

165

occasions in the comprehension analysis. It was stated that more
complex stimuli were expected to be associated with lower driver
comprehension and thus higher accident involvement. Factors affecting
the complexity of the driving environment include weather and road
surface conditions that may demand a larger share of driver attention,
detracting from mental capacity available for the tasks of perceiving
and interpreting signal displays. It would then be expected that
harsher weather and poor road surface conditions may thus be associated
with higher accident involvement ratios, especially for more complex
intersection maneuver tasks, such as turning left. To this end, older
driver accident involvement ratios for various weather and road surface
conditions are investigated in order to assess their impact on older
driver accident performance.

Two additional variables examined in this section allow a better
definition of the older driver "profile;" information on accident type
and driver violation allows the identification of the most hazardous
situations for older drivers under each type of left-turn control.

Finally, involvement ratio differences between sexes under each
type of left-turn control are presented and discussed. Significant
differences in percentages of licensed drivers and driving habits among
sexes were identified both in the literature and during the analysis of
the comprehension database.

VVeather

Older driver daytime accident involvement is lower under adverse
weather conditions (rain and snow--0.95 and 0.86 respectively--table
5.18) than clear weather or fog conditions (1.18 and 1.17 respectively).
A partial explanation for this phenomenon may be found in speculations
by certain authors that older drivers particularly vulnerable under such
conditions may refrain from driving in rain and snow. Thus, only those
among older drivers that are in the best physical condition attempt to
drive during adverse weather conditions. Given their own perceptions of

the conditions, it appears that older drivers assess the danger of

 

 

 

 

166

 

Table 5.18. Older driver accident involvement for different
weather conditions

 

 

 

LEFT-TURN ACCIDENTS

WEATHER ALL

CONDITION DAYTIME

ACCIDENTS PERMITTED PROTECTED PROT/PER

Clear/cloudy 1.18 2.14 1.76 2.49
Fog 1.17 (l) (1) (1)
Rain 0.95 3.59 (1) 2.30
Snow 0.86 2.82 (1) (1)

 

 

 

 

 

 

 

(1) Insufficient data for analysis.

 

driving in adverse weather conditions than when driving in clear
weather. However, despite their overall performance, older drivers are
highly overinvolved in left-turn accidents at permitted locations during
adverse weather conditions (3.59 in rain, 2.82 in snow), an indication
that they may not be aware of the extent of the dangers they face
particularly when turning left at such locations under these conditions.
Their involvement is lowest in clear weather (2.14).

Under protected/permitted phasing, older drivers are less involved
in rainy (2.30) than Clear weather (2.49), but differences are not
dramatic.

Daytime intersection statistics mainly representing the straight-
through maneuver show that Older drivers compensate adequately for
adverse weather conditions when moving straight, but have a higher
involvement during such conditions when turning left at approaches with
permitted control. Diversion of attention to tasks related to turning
left may be responsible for the Observed involvement patterns.

Road surface

Older drivers are less involved in daytime accidents under wet and
snow/ice road conditions than on dry roads (table 5.19). However, they
are significantly more involved in accidents on wet roads when turning
left, regardless of approach left-turn control. Similar trends for
total intersection accidents have been reported by others (125).

Cumulative daytime statistics overwhelmingly represent through

 

 

167

accidents. Thus, Older drivers appear to be able to correctly assess
(even overcompensate) for adverse conditions when performing simpler

intersection maneuvers (i.e., driving straight-through), but misjudge

the risks of getting involved in accidents when executing the more

complex maneuver of turning left under adverse roadway conditions.

 

Table 5.19. Older driver accident involvement for different road
surface conditions

 

 

 

 

 

 

 

LEFT-TURN ACCIDENTS
ROAD ALL DAYTIME
SURFACE ACCIDENTS
CONDITIONS PERMITTED PROTECTED PROT/PER
Dry 1.20 2.15 1.78 2.49
Wet 1.00 2.71 2.88 2.65
Snow or Ice 0.90 2.53 (1) (l)

 

 

 

(1) Insufficient data for analysis.

 

Accident type

Older driver accident involvement in rear-end accidents between
vehicles moving straight is the lowest among accident categories
analyzed in the present study (.85--table 5.20). Involvement in right-
angle accidents between vehicles moving straight is 1.15, and between a
vehicle moving straight and a turning vehicle is 1.36. Finally,
involvement in head-on collisions between a vehicle turning left and one
moving straight is 1.67, and that in rear-end collisions involving

vehicles turning left is 1.78.

 

Table 5.20. Older driver accident involvement for different
accident types

 

 

 

 

 

LEFT-TURN ACCIDENTS
ACCIDENT TYPE ALL
DAYTIME
ACCIDENTS PERMITTED PROTECTED PROT/PER

Angle—straight 1.15 (1) (1) (1)
Rear-end 0.85 (1) (1) (1)
Angle-turn 1.36 1.52 1.09 1.56
Rear-end/left 1.78 2.44 (1) (1)
Head-on/left 1.67 2.47 2.86 3.02

 

 

 

 

(1) Insufficient data for analysis.

 

 

 

 

 

168

The following scenario is a logical explanation of differences in
over-involvement among accident types. Older drivers can quite
effectively keep proper distances from leading vehicles, thus
involvement in rear-end accidents is lowest. However, they fail to
monitor the through signal with the same efficiency, thus violate red
indications and become involved in right-angle (head-on) accidents with
vehicles moving straight (turning left) on other intersection legs.

When turning left they divert their attention to tasks related to the
left-turn maneuver (such as identifying the signal display and
comprehending its meaning, identifying the proper turn lane and path)
thus failing to monitor their distance from leading vehicles and getting
involved in rear-end left-turn accidents. This explains the discrepancy
between a low rear-end accident involvement when no left-turning vehicle
is involved and a high rear-end involvement when a left-turning vehicle
is involved (assuming that the left-turning driver is mostly at-fault).
However, these mechanisms remain hypothetical and further research is
necessary to validate them.

For older driver left-turn accidents at permitted locations,
highest involvement is associated with head-on left-turn accidents
(2.47) and rear-end collisions between vehicles turning left (2.44).
Collisions at an angle involving a left-turning vehicle have an
involvement ratio of 1.52. The previous scenario provides an
explanation for these findings as well. Involvement ratios for
protected and protected/permitted locations, where the highest older
driver involvement is associated with head-on left-turn accidents (3.02
and 2.86 respectively) and lowest with angle-turn accidents (1.56 and
1.09) respectively, are also consistent with the above scenario.

Older drivers can effectively keep their distances from leading
vehicles and are less likely to violate a red indication when moving
straight through than when turning left at an intersection. They are

most likely to violate right-Of-way rules and become involved in head-on

 

 

 

169

collisions when turning left under protected/permitted phasing.

Driver violations

Older drivers are overinvolved in daytime accidents when they are
cited for (listed in order from highest to lowest involvement): improper
turn (2.66--table 5.21), improper lane use (1.69), failure to yield
right-of-way (1.65), following too Closely (0.72), and speeding (0.50).
The three highest ranking violations are associated with highest ranking
intersection problems identified by older drivers in personal surveys
(71): Older drivers stated they have difficulties identifying the proper
turning lane, proper turning path, and locating the left-turn signal.
Thus, while drivers are aware of their deficiencies, they seem unable to
use a compensatory mechanism as they successfully do under other
Circumstances. Older drivers tend to drive conservatively, keeping
longer distances from leading vehicles and not driving fast, and are
less likely than other drivers to be cited for driving too closely to
the leading vehicle or speeding. This pattern is consistent with the

previously identified lower involvement in rear-end accidents.

 

 

 

 

 

 

 

 

 

 

 

Table 5.21. Older driver accident involvement for different driver
violations
LEFT-TURN ACCIDENTS

DRIVER ALL DAYTIME
VIOLATION ACCIDENTS

PERMITTED PROTECTED PROT/PER
Speeding 0.50 (1) (1) (1)
Fail to yield 1.65 2.30 2.09 2.67
Improper lane 1.69 1.64 (l) (1)
Improper turn 2.66 3.32 (1) (1)
Too close 0.72 1.50 (1) (1)
(1) Insufficient data for analysis.

 

For left-turn accidents under permitted phasing, the highest
involvement is associated with improper turns (3.32), failure to yield
right of way (2.30), improper lane use (1.64), and following too closely
(1.50). Permitted left-turn accident patterns are similar to those for

daytime accidents.

170

In summary, older drivers moving straight through are less likely
to be speeding or driving too closely to the leading vehicle. They are
more likely to make improper lane Changes and improper turns or fail to
obey the right-of—way rules. These problems are exacerbated when
turning left at permitted locations. Failure to yield right of way is
more prominent at permitted/protected locations. These findings are
consistent with a higher accident involvement where more complex
situations are present: left-turn accident involvement is higher than
through accident involvement, and violations are highest where right-of-
way rules are most complex, i.e., at permitted/protected locations.

Driver sex

Older female drivers have a slightly higher involvement ratio than
their male counterparts only for total normal signal operations
accidents (1.16 female versus 1.10 male-—table 5.22). They are less
involved in all other accident categories under examination. Thus,

older female drivers, despite driving less miles per year and having

 

Table 5.22. Older driver accident involvement for driver sex

 

 

 

LEFT-TURN ACCIDENTS
DRIVER ALL DAYTIME
SEX ACCIDENTS
PERMITTED PROTECTED PROT/PER
Female 1.16 2.15 1.67 2.12
Male 1.10 2.42 2.11 2.89

 

 

 

 

 

 

 

(1) Insufficient data for analysis.

 

lower correct answer rates (see comprehension analysis), prove to be
more efficient in compensating for situations where they are vulnerable
than their male counterparts.

Accident and driver characteristics: summary and discussion

Older drivers overcompensate (i.e., have a lower involvement
ratio) for adverse weather conditions when moving straight-through, but
are overinvolved in accidents during the same conditions when turning

left at permitted locations. The complexity of the left-turning

171

maneuver and weather interference may be the underlying reason that
Older drivers do not perform as efficiently under these conditions.

Similar patterns are observed for roadway conditions: lowest
involvement in daytime accidents is observed under adverse weather
conditions (wet, snow, or ice) and highest under favorable (dry)
conditions. For left-turn accidents, lowest involvement occurs under
dry pavement conditions and highest under rainy conditions. Drivers may
not able to COpe with adverse roadway conditions when turning because
their attentions are turned to the left-turning tasks.

Older drivers moving straight have a lower involvement ratio in
rear—end accidents than other driver ages. They are also less likely to
get involved in angle, rear-end or head-on accidents with a turning
vehicle. However, they run a higher risk to become involved in the
latter three types of accidents when their attention is consumed with
the left-turning tasks. Involvement in head-on accidents when turning
left is highest at locations with protected/permitted phasing.

Older drivers are less likely than other drivers to be involved in
accidents where they are cited for speeding or following the leading
vehicle too closely when they are moving straight. They have the
highest overinvolvement in accidents where they are cited for improper
lane use, Improper turns and failure to yield. Citation patterns are
consistent with problems older drivers recognize they have within the
intersection environment. Older driver accident problem is exacerbated
when turning left, and failure-to-yield citations are particularly
prominent at permitted/protected locations, where right-Of-way rules are
the most complex among all left-turn phasing types.

Male drivers are underinvolved in total normal signal operations
accidents, however, female drivers are underinvolved in turning
accidents. Female drivers seem to be better able to handle more complex

demands placed on the driver than their male counterparts.

172

5.5. CHAPTER SUMMARY
Approximately four-fifths of intersection accidents involve

vehicles moving straight-through and occur during the daytime. Four-
fifths of turning accidents are related to vehicles turning left and the
remainder to vehicles turning right. Not surprisingly, permitted
phasing violations account for a higher percentage of left-turn
accidents than locations with protected phasing for all driver ages. As
drivers age, the percentage of through accidents decreases while the
percentage of turn accidents (both left and right turns) increases.

Older drivers are overinvolved in intersection accidents. Their
overinvolvement in turning accidents is significantly more pronounced
than that in accidents involving vehicles moving straight. Among
turning accidents, involvement is highest for left turns and less
pronounced for right turns. Older driver overinvolvement in left-turn
accidents depends on approach left-turn control. The highest
involvement is observed at permitted/protected and the lowest at
protected locations.

Differences in older driver involvement ratios among intersection
maneuvers can be explained based on intersection maneuver complexity.
The simplest maneuver, driving straight through places minimal
requirements on the driver, since the driver keeps driving in the same
lane and only needs to monitor the leading vehicle and the through
signal display. Through signal intervals have standard meanings and are
easily interpreted. Thus, the through accident involvement ratio is
lowest for older drivers. Drivers turning right may need to change
lanes and monitor pedestrian traffic crossing the side street in
addition to monitoring leading vehicles and signal indications. The
higher complexity of this maneuver is reflected in higher older driver
involvement ratios in right-turn accidents. Finally, the left-turn
maneuver involves all tasks associated with right turns (with the
difference that pedestrians crossing the cross street are further from

the driver and thus harder to monitor) and the additional task of

173

monitoring opposing traffic in order to identify acceptable gaps. Older
drivers have been demonstrated to have significant difficulty correctly
judging speeds of oncoming traffic and/or acceptable gaps. Furthermore,
the comprehension analysis indicated that older drivers have a
particularly low comprehension of left-turn signal indications. The
complexity of the left-turn maneuver is reflected in higher Older driver
involvement ratios in left-turn accidents.

Protected left-turns relieve the driver of the task of monitoring
opposing traffic. In addition, the protected left-turn green arrow has
a unique meaning. Thus, involvement ratios for protected left-turns are
the lowest among alternative left—turn phase types. Left-turns under
permitted phasing require the driver to monitor opposing traffic. The
double use of permitted green balls as an indication granting the right-
of-way to through traffic and an indication for permitted left-turns may
also lead to driver confusion. The higher complexity of the tasks
required under permitted phasing are reflected in higher involvement
ratios for these locations. Finally, protected/permitted phasing
provides the most complex left-turn maneuver since right-of-way rules
Change during the cycle and signal displays are Often more complex with
the presence of multiple illuminated sections. Thus, older driver
involvement ratios under this phasing are the highest among all left-
turn phasing types examined.

A number of signal, geometric, and accident variables were found
to be related to older driver accident involvement. The most
significant findings are summarized here.

Permitted phase duration was found to be directly related to left-
turn accident involvement. Longer permitted phase durations are
associated with a higher number of vehicular conflicts and more
opportunity for accidents. Similar patterns were identified for
protected phase duration at permitted/protected locations. Longer
protected phases at such locations are associated with heavier left-turn

demand (the protected interval is used to accommodate vehicles that did

174

not have an opportunity to turn during the permitted phase). Thus, the
protected phase duration is a likely surrogate for the real cause of
accident overinvolvement, i.e., a heavy left-turn demand in the presence
of the most complex right-of-way rules among all left-turn phasing
strategies. Drivers are under pressure to complete left turns while
continuously monitoring the left-turn signal for current right-of-way
rules. The presence of an additional change interval (from protected to
permitted or vice—versa) provides more opportunity for conflicts than
any other phasing strategy. The complexity of the situation affects
Older drivers to a higher degree than their younger counterparts as
explained previously. Therefore this finding can be explained in terms
of opportunity of vehicular conflicts as well.

Simpler (stacked-three) left-turn signal face arrangements used at
protected and protected/permitted locations are associated with lower
accident involvement ratios than stacked-four displays. This finding is
in agreement with comprehension analysis findings, where older drivers
were found to comprehend simpler displays better.

Involvement ratios at locations with leading protected left-turn
intervals were found to be lower than those at lagging intervals both
under protected and protected/permitted phasing. Inferior lagging phase
performance was attributed to opposing vehicles violating the change
interval between the permitted and lagging protected phases.

Nighttime involvement ratios were lower for locations with
flashing yellow balls than for those with flashing red balls. These
findings correlate with comprehension analysis results indicating that
serious error rates (see methodology for definition) among older drivers
are lowest for flashing yellow balls and highest for flashing red balls.
Locations with 24-hour full color operations have the highest nighttime
involvement ratios. This is thought to be a direct consequence of the
presence of higher volumes at such locations putting older drivers at a
distinct disadvantage since they are more susceptible to glare and have

longer glare recovery times than their younger counterparts, thus an

175

increase in opposing traffic volumes may have a disproportionately
adverse effect on older driver performance.

The presence of left-turn lanes is associated with lower
involvement in all daytime and left-turn accidents regardless of left—
turn phase type. This is likely due to the effective elimination of
interference between left and through traffic.

Left-turn signals placed closer to the driver were found to be
associated with lower left-turn accident involvement ratios. A closer
placement allows for more efficient scanning for, and identification of,
the left-turn signal. This is particularly important for older drivers
that have narrow peripheral vision: the larger the angle between the
signal and the vehicle path, the higher the Chances that they will not
be able to simultaneously scan the left-turn signal as well as monitor
leading and opposing vehicles and, thus, the higher the likelihood of
accident involvement.

Although older drivers seem to be able to adequately compensate
for adverse weather and roadway conditions when moving straight, they
are overinvolved in accidents when turning left under adverse
conditions. The tasks associated with left turns compared to those for
going "through" consume a disproportionate share of drivers' attention,
thus they tend to be more vulnerable to accidents.

Older drivers are overinvolved in rear-end, head-on, and angle
accidents while turning left, regardless of approach left-turn control,
than when moving straight. A possible explanation for these occurrences
is that left-turning tasks detract from their ability to monitor leading
vehicles and/or monitor signal indications.

Based on the literature, older drivers are aware of their major
problems in the intersection environment such as the inability to locate
signals, proper turn lane, and turning path, and have accurately
reported them in personal surveys. However, they seem to be unable to
compensate for these deficiencies as witnessed by a preponderance of

Citations for failure to yield, improper lane use, and improper turns.

176

Overinvolvement in accidents where such citations were issued is more
prominent when turning left at locations with permitted phasing.

In summary, older drivers are generally better able to cope with
driving tasks in simpler, rather than more complex situations. Although
they seem to be aware of hazardous situations and effectively cope with
certain adverse conditions (such as driving in bad weather) even better
than other drivers, they are able to do so only when performing simpler
tasks (e.g., driving straight). They appear to become overwhelmed as
soon as the driving task becomes more complex-for example, turning left
in bad weather when involvement becomes progressively higher for

protected, permitted, and protected/permitted locations.

CHAPTER 6
SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS
6.1. INTRODUCTION
This research was an attempt to gain insights into the morphology
and causes of older driver left-turn accident experience at signalized
intersections. The basic hypothesis underlying this endeavor is that
older driver accident overinvolvement is due to a reduced ability to
make correct decisions in complex driving situations. The effect of
various forms of driving environment complexity on older drivers was
examined in two independent analyses:

i) A driver comprehension analysis based on laboratory data
gathered during a previous study of left-turn signal
displays sponsored by the FHWA and,

ii) An independent analysis of accident experience based on
actual accident, signal, and geometric data specifically
gathered for the present study.

The laboratory analysis was directed to measuring effects of left-

turn signal display complexity using Older driver comprehension as a
measure of effectiveness. The field data analysis was also addressed to
signal display complexity but was primarily concentrated on intersection
maneuver complexity; disproportionate involvement in accidents by Older
drivers was used as a measure of effectiveness for this analysis. An
effort was maintained throughout to examine "parallel" variables in the
laboratory and field analyses and test the soundness of the hypothesis
that signal display characteristics associated with reduced older driver
comprehension measured in the laboratory may be related to
disproportionately higher accident experience in the field.

In addition to variables pertaining to driving environment

complexity, driver information variables included in the laboratory
database, and signal, geometric and accident characteristics variables

from the field database were analyzed.

177

 

 

178

6.2. COMPREHENSION ANALYSIS SUMMARY

The analysis of driver information variables collected as part of
the comprehension database verified that trends of reduced miles driven
per year as drivers age, identified in the National Personal
Transportation Survey, were present in the subject sample. Thus, older
subjects were found to drive less than any other age group and older
female subjects were found to drive less than their male counterparts.

The database provided insights to driver-specific factors related
to older driver comprehension of left-turn stimuli. Correct
comprehension of left-turn signal displays was found to be related to
miles driven per year--older drivers (that drive less than other age
groups) are more likely to misinterpret such displays. Mitigating this
finding is the observation that misinterpretations are more often than
not associated with older drivers giving away their right of way
(committing a minor error) rather than violating another driver's right
of way (committing a serious error). Although serious errors are
infrequent, older drivers commit significantly more such errors than
other drivers; however, no significant differences exist between sexes.
Interpretation of left-turn stimuli displays is related to number of
years of driving experience: correct interpretation increases with
experience to a point and then declines as detrimental aging effects
appear to offset beneficial experience effects.

Inter-interval analysis

The driver comprehension analysis based on stimulus message showed
that older driver left-turn signal misinterpretations leading to
violations Of other drivers' right of way are most likely when facing
flashing and permitted left-turn displays, less likely when facing red
displays, and least likely when facing Change interval displays.

Stimuli used exclusively to demand the most urgent driver reactions
(such as preparation to stop for a change interval or red ball) are

least likely to be misinterpreted; those requiring interpretation (such

 

 

179

as flashing nighttime red ball stimuli whose meaning depends on the
simultaneous through indication) are most likely to be misinterpreted.
Thus, higher complexity stimuli are related to a higher incidence of
older driver serious errors.

lntra-interval analysis summary

The most important findings based on comparisons of alternative
left-turn displays conveying similar messages (e.g., alternative
displays used for permitted left-turns) are summarized in the following
paragraphs.

Permitted interval

Older drivers were found to be less likely to violate other
drivers' right of way due to misinterpretations of the permitted
interval in situations where the lens illuminated on the left—turn
display is not a green ball (i.e., the left-turn indication is a
flashing yellow or red ball). The higher likelihood of older drivers
violating other drivers’ right of way when both the through and left-
turn indications are green balls is attributed to confusion about the
message of a green ball: the same signal indication grants the right of
way to through drivers, while drivers turning left are expected to
interpret the indication differently. Flashing yellow and red ball
indications are automatically differentiated from through green balls
and are always associated with a measure of caution, thus simplifying
mental processing and improving the chance of correct interpretation (or
at least decreasing the probability of drivers incorrectly presuming
they have the right of way). It is important to note that older drivers
were not found to be more likely than other age groups to incorrectly
think they have the right of way under permitted phasing. However,
older drivers are less likely than other age groups to correctly

interpret permitted indications.

180

Protected interval

Since drivers have the right of way during the protected interval,
misinterpretations of protected displays indicate that drivers
incorrectly think they have to give up the right of way to other
intersection movements. Older drivers are more likely than other ages
to misinterpret the meaning of protected displays.

Misinterpretations were found to be less likely in the presence of the
sign "left-turn signal." The rate of misinterpretation was also found
to be related to illuminated section configuration. Stimuli displaying
green indications (i.e., green arrow alone or simultaneously with green
ball) on the left-turn displays were found to be better comprehended
than stimuli simultaneously displaying a green arrow and red ball.
Better comprehension was found to be associated with a through red ball
than a through green ball. These findings are consistent with the
hypothesis of decreased comprehension for higher complexity stimuli:
When lenses on the left-turn signal are green, the required
interpretation is much less involved than that necessary when
contradictory red ball and green arrow indications are present on the
left-turn signal. Lower comprehension in the presence of a through
green ball may be attributed to additional interpretation required to
differentiate such displays from permitted displays: the driver needs to
make the distinction between simultaneously illuminated through green
ball and left-turn green arrow (protected) and through and left-turn
green ball (permitted).

Red interval

Older drivers are more likely than other drivers to misinterpret a
red left-turn indication. However, they are less likely to misinterpret
red indications placed between the left-turn and the adjacent through
lane than those placed along the left—turn lane centerline. Signal

position affects the permitted green ball in an exactly opposite

 

 

 

 

 

 

181

direction: permitted displays placed in the straight-ahead position are
less well comprehended than those placed between lanes. Both findings
are interpreted as an effect of driver confusion about the dual meaning
of green ball displays: green balls placed between lanes (closer to the
through signal) lead left-turning drivers to believe that they have the
right of way just like through drivers do. Permitted indications placed
in the straight-ahead position (further from the through signal) are
less likely to be confused with the through signal message and are
therefore better comprehended. It is possible that a green ball carry-
over effect may lead drivers facing a red ball on the left-turn signal
to treat it differently than a through red ball when the signal is
placed in the straight-ahead position (further from the through signal)

than when it is placed between lanes (closer to the through signal).

Change interval

Older drivers are no more likely than other age groups to
incorrectly think they have the right of way during a change interval
(i.e., a yellow indication). However, older drivers are less likely
than other age groups to correctly interpret change interval displays.
Illuminated left-turn and through signal lens configuration were found
to have a significant impact on correct stimulus interpretation.
Comprehension is higher when a yellow arrow or ball is illuminated and
lower when an additional red or green ball is illuminated. Red through
indications enhance, and green indications have an adverse impact on
comprehension.

Simple stimuli are better comprehended than more complex stimuli:
a single illuminated yellow indication does not require extensive
interpretation, since it has a unique meaning. When a different color
lens is also illuminated, additional interpretation is required and
confusion potential increases, thus comprehension deteriorates. A
through red indication reinforces the message of a yellow left-turn

indication, since both convey unique messages urging the driver to stop.

 

 

182

A through green ball may mislead drivers into thinking that the yellow
left-turn indication will be followed by a permitted one, thus they need
not necessarily stop.

Flashing Operation

Older drivers are more likely to violate other drivers' right of
way but not to a significant extent when facing signals under flashing
operation. Older drivers are less likely to violate other drivers'
right of way when facing flashing yellow left-turn indications than when
facing red flashing indications. The lower performance under flashing
red indications is directly related to the higher complexity of such
displays: a flashing yellow left-turn ball is always associated with a
permitted left-turn; the meaning of a flashing red ball can only be
decided in conjunction with the through indication.

In summary, the comprehension analysis demonstrated that:

i) Older drivers have consistently lower comprehension scores
than other drivers. Comprehension differences are more
likely to be significant when measured in percent drivers
correctly interpreting stimuli and less likely to be
significant when measured in percent drivers incorrectly
thinking they have the right of way when they don’t.

ii) Stimulus complexity adversely impacts Older driver
comprehension regardless of stimulus message.

iii) Older driver stimulus comprehension depends on stimulus
message. In terms of serious errors, change stimuli are
best comprehended (fewer errors), followed by red, permitted
and flashing operations stimuli (most errors). Most correct
answers were associated with red interval stimuli followed
by change, protected, permitted and flashing Operations

stimuli (least correct answers).

 

 

183

6.3. FIELD DATA ANALYSIS SUMMARY

The majority of intersection accidents involved vehicles moving
straight through the intersection and occurred during daytime. Turning
accidents most frequently involve vehicles turning left. Left-turn
accidents occur more frequently at locations with permitted phasing than
locations with protected phasing. As drivers age, involvement in
through accidents decreases and a shift to turn accidents is observed.

Older drivers are overinvolved in intersection accidents,
especially when turning. Overinvolvement is highest when executing
left-turns and less pronounced for right-turns. Left-turn control was
found to be related to Older driver left-turn accident overinvolvement
which is the highest at permitted/protected and the lowest at protected
locations.

Maneuver analysis

A possible explanation for differences in older driver
involvement ratios among intersection maneuvers can be found in
intersection maneuver complexity. Simple maneuvers, such as driving
straight through, place minimal requirements on the driver and are
related to the lowest intersection accident involvement ratio. Turning
right may require lane changes and monitoring pedestrian traffic in
addition to tasks normally required for driving straight-through the
intersection. The higher complexity of this maneuver is reflected in
higher older driver involvement ratios in right-turn accidents.

Turning left requires all tasks associated with right turns
(monitoring pedestrian traffic is, however, more difficult since
pedestrian paths are further from the driver) with the additional task
of monitoring Opposing traffic in order to identify acceptable gaps in
pedestrian and vehicular traffic. Furthermore, it was demonstrated that
older drivers have a particularly low comprehension of left-turn signal
indications. Thus, turning left is the most complex intersection task

among the ones analyzed. This is reflected in Older driver involvement

 

 

 

 

184

ratios which are the highest among the maneuvers examined in this study.
Left-turn accident analysis

Among left-turn phasing strategies, protected turns require
simpler tasks than permitted or protected/permitted, since drivers do
not need to monitor opposing traffic,and the green arrow associated with
such turns has a unique meaning. Thus, involvement ratios for protected
left-turns are the lowest among alternative left-turn phase types.
Permitted phasing requires the driver to monitor opposing traffic, and
the double use of permitted green balls as an indication granting the
right of way to through traffic and an indication for permitted left-
turns may lead to driver confusion. The higher task complexity under
permitted phasing is reflected in higher involvement ratios at permitted
locations. Finally, protected/permitted phasing requires more driver
interpretation than any other phasing since right-Of-way rules change
during the cycle, and that, in the presence of more complex signal
displays. Thus, older driver involvement ratios under this phasing are
the highest among all left-turn phasing types examined in the present
study.

Involvement related tO signal, geometric, and accident attributes

A number of signal, geometric and accident variables were found to
be related to older driver accident involvement. The most significant
findings are summarized here.

Longer permitted phase durations were found to be associated with
higher older driver involvement ratios, possibly due to more opportunity
for vehicular conflicts during this phase. Protected phase duration at
protected/permitted locations was also found to be directly related with
left-turn involvement ratios. It was suggested that protected phase
duration may be a surrogate for heavy left-turn demand during the
permitted interval. Therefore it is possible that this finding can be

explained in terms of opportunity of vehicular conflicts as well.

 

 

 

185

Simpler left-turn signal face arrangements (e.g., stacked-three)
used at protected and protected/permitted locations are associated with
lower accident involvement ratios than stacked-four displays. This
finding is consistent with comprehension analysis findings, where older
drivers were found to comprehend simpler displays better.

Leading protected left-turn intervals were associated with lower
involvement ratios than lagging intervals both at locations with
protected and protected/permitted phasing. Inferior lagging phase
performance could be due to opposing vehicles violating the change
interval between the permitted and lagging protected phases.

Flashing yellow balls are associated with lower nighttime accident
involvement ratios than flashing red balls. These findings are in
agreement with comprehension analysis results showing the lowest
comprehension for flashing red balls and highest for flashing yellow
balls. The highest nighttime involvement ratios were found at locations
with twenty four-hour full color operations. It is possible that higher
involvement ratios at such locations are due to higher opposing volumes
causing disproportionately more serious glare problems for older drivers
while simultaneously allowing fewer opportunities to turn (i.e., fewer
acceptable gaps).

Lower involvement ratios were found at locations with exclusive
left-turn lanes. Effective elimination of interference between left and
through traffic is a possible explanation for this finding.

Lower left-turn accident involvement ratios were found where the
left turn signal was placed closer to the driver. It is hypothesized
that a Closer placement allows for more efficient scanning of the left-
turn signal while simultaneously monitoring other turning tasks.

Older drivers are not overinvolved in adverse weather and roadway
conditions when moving straight, but are more vulnerable when turning
left under such conditions. It is possible that left-turning tasks

consume a disproportionate share of drivers' attention and older drivers

 

 

 

186

become overwhelmed with the additional difficulties of lower visibility
and extra caution necessary to deal with adverse environmental
conditions.

Rear—end, head-on and angle Older driver accident overinvolvement
is much more pronounced when turning left than when moving straight,
possibly because left-turning tasks detract from the ability to monitor
leading vehicles and through signal indications.

Older drivers involved in accidents are less likely to be cited
for following too Closely or speeding, but are more likely to be cited
for failure to yield the right of way, improper lane use, and improper
turns. The most common types of citations are most often issued at
locations with permitted phasing.

Overall, older driver accident overinvolvement is less serious in
situations requiring the performance of simpler tasks. Although older
drivers seem to be aware of the dangers involved with certain adverse
conditions (such as driving in bad weather) and cope with them even
better than other drivers, they are only able to do so when performing
simpler maneuvers (e.g., driving straight) under such conditions.
However, they are likely to become overwhelmed as maneuver complexity,
adverse environmental conditions, and signal display complexity
increase.

6.4. CONCLUSIONS

The laboratory analysis of older driver comprehension of left-turn
displays demonstrated that Older drivers have lower left-turn display
comprehension scores than other drivers. Laboratory results indicate
that incorrect signal interpretations among all drivers are more likely
to be related to a choice to yield when they have the right of way
rather than violate other drivers' right of way. Although violations of
other drivers' right of way are quite infrequent, significant
differences exist between Older and other drivers.

Incorrect interpretations of left-turn signal displays among older

 

 

 

187

drivers depend primarily on stimulus message and display complexity.
Stimuli used to convey a single, simple message, and especially those
conveying the most urgent messages (such as yellow ball or red ball
conveying the messages prepare to stop and stop respectively) are better
comprehended than those used to convey multiple messages (e.g., green
ball, flashing operations left-turn red ball). Simpler stimuli using a
single illuminated lens on the left-turn signal are better comprehended
than those using two lenses.

Expectations that the presence of higher complexity displays
associated with lower older driver comprehension would adversely affect
older driver accident involvement were verified with field data.

Indeed, Older driver accident involvement was higher at locations with
stacked four-section left-turn displays with two simultaneously
illuminated sections than at locations with the better comprehended
stacked three-section displays with a single illuminated section;
flashing operation red ball indications with higher serious
comprehension errors than flashing yellow ball indications were present
at locations with higher left-turn accident involvement.

Intersection accident statistics roughly reflect the distribution
of volumes among the through, left-, and right-turn maneuvers. Thus,
almost four-fifths of at-fault drivers involved in intersection
accidents are driving straight through with the remainder involving
drivers turning left or right. A stratification of accidents according
to approach left-turn control reveals that left-turn accident percentage
differences exist based on the approach’s left-turn control. As drivers
age, the percentage Of accidents involving a turning maneuver increases
(percentage of through accidents decreases), especially for left-turns.
Approaches with protected left-turn control have the lowest left-turn
accident percentage, and those with permitted left-turn control the
highest for each driver age group.

Older drivers are overinvolved in intersection accidents. The

 

 

 

 

 

 

188

degree of overinvolvement depends on intersection maneuver: it is lower
for drivers moving straight through the intersection, higher for drivers
attempting a right turn, and highest for drivers turning left.

Increased overinvolvement correlates with additional tasks required for
turning maneuvers compared to through movements (e.g., monitoring
pedestrian traffic and/or Opposing traffic).

Among left-turn accidents, overinvolvement is lowest under the
simplest right of way rules, i.e., at locations with protected phasing,
where drivers have the right of way over all other conflicting
intersection movements. Left—turn accident involvement is highest at
locations with permitted/protected phasing which is associated with the
most complex right of way rules under which drivers have the right of
way or have to yield to Opposing traffic depending on the signal
indication. Turning left under permitted/protected phasing is
complicated by the need to continuously monitor the left-turn signal
display and opposing traffic and correctly interpret the right-of-way
rules at any instant.

Older driver susceptibility to environment complexity is also
demonstrated by higher accident involvement in situations where
opportunities for conflict with other vehicular movements are elevated
(such as at approaches where no left-turn lanes are provided, or where
permitted phase durations are longer), where more intense scanning is
required in order to locate the left-turn signal (e.g., when the left-
turn signal is located further from the driver), or where the
difficulties of an already complicated driving task (such as turning
left) are compounded by the presence of adverse meteorological
conditions.

The older driver "profile" is one of a driver who drives fewer
miles per year, avoids nighttime and adverse weather driving, does not
exceed the speed limit, keeps adequate distances from leading vehicles

and, when unsure about the meaning of a signal display, and is more

 

 

189

likely to grant the right of way to other drivers than violate their
right of way. This finding based on laboratory results under simplified
conditions in which older subjects could clearly see stimuli and had
adequate time for mentally processing their messages are reversed in the
field. Older drivers are more frequently cited for right of way
violations when involved in accidents, perhaps, because they cannot
clearly see left-turn displays and do not have adequate time to process
signal information given that driving tasks in the field are far more
complex than the laboratory environment.

Although older drivers seem to be able to correctly assess their
risks under certain circumstances and effectively avoid being
overinvolved in accidents (e.g., when driving straight during daytime,
even in adverse weather conditions), they are unable to perform more
complex tasks with the same efficiency and become overinvolved in
accidents (e.g., turning left under adverse weather conditions).

Hypotheses based on older subject comprehension measured in the
laboratory were generally verified using field data. Older driver poor
performance measured in comprehension in the laboratory and accident
experience in the field can be related to left-turn display and/or field
parameter complexity.

6 . 5 . RECOMMENDATIONS

The present research covered a broad spectrum Of variables
associated with older driver comprehension of left-turn displays and
older driver involvement in left-turn accidents. However,
not all variables investigated and found to be related to driver
comprehension and/or accident involvement can be readily modified in the
field in order to improve safety (e.g., longer permitted phases found to
be related to higher older driver left-turn accident involvement cannot
be altered at will). Thus, the following recommendations are
necessarily limited to variables with a practical significance for the

practitioner, that is, geometric and signal features variables that can

 

   

190

be reasonably expected to be amenable to adjustment in the field.

The safety benefits of simpler driving environments for older
drivers have been demonstrated both in terms of better comprehension of
left-turn signal displays and in terms of reduced accident involvement.
The following recommendations based on the data analyzed herein are made
with regard to left-turn signal displays:

Based on the comprehension analysis results:

i) The use of permitted left-turn green ball indications should be
avoided in favor of indications using yellow or red flashing
indications. Permitted left-turn green ball indications are less
well comprehended and it is believed that driver confusion arises
from the dual meaning of left-turn green ball sections that grant
the right of way to through drivers but are expected to be
interpreted differently from left-turning drivers. Better
comprehension Of the suggested alternative displays is believed to
be due to their unique meanings and their differentiation from
through signal indications that eliminate driver confusion to a
significant extent.

Based on the accident analysis results:

i) Construction of a left—turn lane at approaches with permitted or
protected/permitted left-turn control is recommended as a measure
to reduce older driver overinvolvement in left-turn accidents.

ii) It is recommended that older drivers be made aware that they may
not be able to correctly assess their vulnerability to left-turn
accidents in adverse weather conditions and on roads that are wet
or covered with snow. They should be advised to avoid driving
under such conditions, but, should they have to, they should
carefully Choose alternate routes that will avoid left-turns as
much as possible.

Based on both comprehension and accident analysis results:

i) Permitted displays be placed along the extension of the left-turn
lane centerline instead of between the left-turn and the adjacent
through lane. Placement in the straight-ahead position concurs
with accident analysis results pointing out that older driver
left-turn accident overinvolvement is lower at locations where the
left-turn signal is placed Closer to the driver. This
recommendation could be implemented as a first, less expensive
step in a permitted left-turn safety improvement program, before
signal sections and controllers are equipped to operate under a
flashing red or yellow permitted mode as recommended above.

ii) The use of stacked four-section displays with (fourth level) green
arrow simultaneously illuminated with a green ball should be
abandoned in favor of stacked three-section displays with a red
ball, yellow ball and green arrow sections. A green arrow
illuminated section should be used at approaches with protected
left-turn phasing and green arrow/flashing red ball (for the
protected and permitted phases respectively) should be used at
approaches with protected/permitted phasing. These

 

 

191

recommendations are based both on the comprehension analysis
(lower comprehension was found for stacked four-section displays)
and the accident data analysis (higher involvement at approaches
with four-section displays).

The following recommendations are based on preliminary results;
further investigation is necessary before they can be adopted:
Based on the accident analysis:

i) Leading protected left—turn phasing be used at locations with
protected or protected/permitted left-turn phasing. A lower
performance of locations using lagging protected phasing was
attributed to opposing through drivers violating the change
interval preceding the protected phase. However, further research
is recommended in order to identify the underlying causes for the
observed differences in accident experience between leading and
lagging protected phasing.

ii) It was found that older driver nighttime accident involvement was
higher at locations with 24-hour full color operations. However,
before a recommendation is made about using flashing operations
during nighttime, it is recommended that additional data be
collected and analyzed with regard to the merits of flashing
versus 24-hour signal operation. This is because there were
factors other than signal operation mode (e.g., glare recovery
time that may disproportionately affect older drivers in the
presence of higher nighttime traffic volumes) that may have a
stronger influence on accident experience.

iii) It is recommended that further study of alternatives to making
left-turns (e.g., turning right around the block) be conducted and
their merits be compared to making left-turns before older drivers
are advised about specific alternatives to making left turns.

The findings of the laboratory comprehension analysis pertaining
to particular signal phases can be integrated with subsequent efforts
dedicated to the study of driver behavior during particular signal
phases in the field. Traffic conflict studies, for example, may reveal
which phases and other signal characteristics are associated with the
most dangerous and/or frequent violations of right of way rules, thus
weights could be assigned to the importance of correctly comprehending
each particular phase. Accident countermeasures could then be fine-

tuned to address problems with particular phases.

A summary of the above recommendations is presented in table 6.1.

 

 

 

 

 

192

 

aumnmmumc
noummnou nonuusm

mammamcm ucmpduod

.mcoauuwmumucq um uuma acacuzu
on m0>wumcumuam Omuumwmum co mum>wuo umOHO mumosum

 

hummmmumc
noumwmmu umnuuah

namhamcm ucmpwoom

.manammom mums: Tawuuzmac ocwusp cowumummo maficmmHu
. no uo>mu ca COwumummo Hmcmam usoniem mumcwswam

 

Nummmmuoc
consumes umnuusm

mamhamcm
ucmpwoofi

.manammmm
Oumn3 manna cusuluwma Omuumuoum mcapmma umwmum

 

mamaamcm cowncmzmumsou

.mNmHmmwc

 

 

 

 

 

09030000 new ucmpwuod mmunulpmxumum :uw3 mamammwc uDOwipmxumum mumHmmm
man>amcm Seamcmsmumeoo .mcwaumucmu Dana cusuluuma

mumavmnc 0cm ucmpwuod 0:» mo SOsmcmuxm 0:» macaw namcmwn cucuiuuma Conan
.ECOwuHOcoo

umcumw3 mmu0>pm umpcs >Hamwommmm .uuma mcwcusu

mumswmpm manaamcm ucmpwoum c023 xnwu penumuuca usonm mum>wup umvao Oumospu
.mcowumooH Owuu45u0m\pmuumuoum

mumnvmvd mamhamcm ucmuwuoa 0cm cmu048umm um mmcma cusuiuuwa m>wmsaux0 mpw>oum
.mCOwumoapcw own no 3O~Hm> ocwnmmau

mumsvmpm OADSHmcm codmcmnmumeoo nu«3 mc04umuwpca damn cumum Omuuaeumm mumHmmm
even :0 Conan sowvausoaaooom

 

 

 

 

neodumpcmaeoomu mo mumfifism H.0 manna

 

 

 

 

 

APPENDIX A

 

 

 

APPENDIX A

Comprehension analysis stimuli

Information in appendix A is based on the research project
entitled "Signal Displays For Left Turn Control" (U.S. DOT Contract No.
DTFH 61-85-C—00164). Details can be found in Task B report: ”Driver
Understanding of Signal Displays" prepared by Mark Freedman and David P.
Gilfillan of KETRON, INC. for JHK & Associates in April, 1988.

The term "signal face arrangement" is used herein to indicate
characteristics of signal heads (such as number and position of signals,
traffic movements they address, signal face composition), and presence
of left turn signs and left turn lanes. For a typical signal face
arrangement refer to figure 3.2. The term "stimulus" is used to
indicate a particular combination of illuminated signal lenses during a
signal interval. The same signal face arrangement may be used for a
number of stimuli (i.e., signal intervals).

Stimuli numbers used in the present appendix correspond to those
used in the report mentioned above. In all 17 signal face arrangements
used as a base for 81 stimuli are summarized in the following tables.

Information on number and position of left turn and through signal
heads, presence of supplemental signs, and presence of exclusive left
turn lane is shared by all stimuli based on a signal face arrangement.
Each stimulus is represented by a schematic of one left-turn and one
through signal face, where lens types/functions are indicated by the
following abbreviations:

RB: Red Ball

RA: Red Arrow

FRB: Flashing Red Ball
FRA: Flashing Red Arrow
YB: Yellow Ball

YA: Yellow Arrow

FYB: Flashing Yellow Ball
FYA: Flashing Yellow Arrow
GB: Green Ball

GA: Green Arrow

FFGA: Fast Flashing Green Arrow
BlankzLens Not Illuminated

193

 

 

 

1594

Both full color and flashing operations stimuli are present in the
database and are identified as such. Certain flashing operations left—
turn signal faces remain dark, thus no abbreviation is shown on the

left-turn signal schematic for such stimuli.

 

Table A1 Permitted stimuli
Signal face arrangement No 1

 

 

 

Left turn signal characteristics Through signal characteristics
Nunber of signal heads: 1 Nurber of signal heads: 1

Vertical position: Pedestal Vertical position: Pedestal
Horizontal position: Far left curb Horizontal position: Far right curb
Supplemental sign: None Special notes: Single-lane approach

 

Full color operations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

R8 RB
68 GB
Left Through Left Through
Stimulus NO 1 Stimulus NO 2

 

Signal face arrangement No 2

 

 

Left turn signal characteristics Through signal characteristics

Number Of signal heads: 1 Number of signal heads:1

Vertical position: Overhead Vertical position: Overhead

Horizontal position: Between lanes Horizontal position: Over through lanes
Supplemental sign: None Special notes:Exclusive left turn lane

 

Full color operations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

R8 RB
68 GB
Left Through Left Through
Stimulus NO 3 Stimulus NO 4
Flashing Operations
FRB FRB
FYB FYB
Left Through Left Through

Stimulus NO 5 Stimulus No 6

 

 

 

 

 

 

 

 

1595

 

Table A2 Protected stimuli

 

Signal face arrangement NO 3

 

Left turn signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Left-turn lane
centerline

Supplemental sign: Left turn signal

Through signal characteristics

Number of signal heads: 2

Vertical position: Overhead

Horizontal position: Over through lanes
Special notes: Exclusive left turn lane

 

 

Full color Operations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

RB RB R8 R8
YB
GA
Left Through Left Through Left Through
Stimulus NO 7 Stimulus NO 8 Stimulus No 9
Flashing Operations
FRB FRB FRB FRB
FYB FYB FYB FYB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Left Through Left Through Left

Through Left Through Left Through

Stimulus No 10 Stimulus NO 11 Stimulus No 12 Stimulus NO 13 Stimulus No 14

 

Signal face arrangement No 4

 

Left turn signal characteristics

Number Of signal heads: 1

Vertical position: Pedestal

Horizontal position: Far median between lanes
Supplemental sign: Left turn signal

Through signal characteristics

Number Of signal heads: 2

Vertical position: Overhead

Horizontal position: Over through lanes
Special notes: Exclusive left turn lane

 

 

Full color operations

 

 

 

 

 

R8 R8 R8

 

 

 

 

 

 

 

GA

 

 

 

 

 

 

 

 

 

 

 

Left Through Left Through
Stimulus NO 15 Stimulus NO 16

 

Flashing operations

 

 

 

 

 

 

 

 

FRB

 

 

 

FRB FRB FRB

 

 

 

 

FYB FYB FYB

 

 

 

FYB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Left Through Left Through Left

 

Through Left Through Left Through

Stimulus NO 17 Stimulus No 18 Stimulus No 19 Stimulus NO 20 Stimulus No 21

 

 

V

1596

 

Table A2 (Cont’d).

 

Signal face arrangement NO 5

 

Left turn signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Left-turn lane
centerline

Supplemental sign: Left turn signal

 

Through signal characteristics

Number of signal heads: 2

Vertical position: Overhead

Horizontal position: Over through lanes
Special notes: Exclusive left turn lane

 

Full color operations

 

 

 

 

 

YA

 

 

 

 

 

 

 

 

Left Through
Stimulus NO 22

 

Flashing Operations

 

 

 

 

 

FYA FYB

 

 

 

 

 

 

 

 

Left Through
Stimulus No 23

 

Signal face arrangement No 6

 

Left turn signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Left-turn lane
centerline

Supplemental sign: Left turn signal

 

Through signal characteristics

Number Of signal heads: 2

Vertical position: Overhead

Horizontal position: Over through lanes
Special notes: Exclusive left turn lane

 

Full color operations

 

 

 

RA RB

 

 

 

 

 

 

 

 

 

 

Left Through
Stimulus NO 24

 

Flashing Operations

 

 

 

 

 

FRA FRA FRB

 

 

 

FYB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Left Through Left Through
Stimulus No 25 Stimulus NO 26

 

 

 

 

1197

 

Table A2 (Cont'd).

 

Signal face arrangement No 7

 

Left turn signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Left-turn lane
centerline

Supplemental sign: None

Through signal characteristics

Number Of signal heads: 2

Vertical position: Overhead

Horizontal position: Over through lanes
Special notes: Exclusive left turn lane

 

 

Full color Operations

 

 

 

 

 

 

RB RB RB

 

 

 

 

 

YB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

GA

 

 

 

 

 

 

 

Left Through Left Through Left
Stimulus NO 27 Stimulus NO 28 Stimulus No 29

Through

 

Flashing operations

 

 

 

 

 

 

 

 

 

 

FRB

 

FRB FRB FRB

 

 

 

 

 

 

 

 

FYB FYB FYB

 

 

FYB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Left Through Left Through Left

Through Left Through Left Through

Stimulus NO 30 Stimulus NO 31 Stimulus No 32 Stimulus NO 33 Stimulus NO 34

 

 

 

 

 

 

 

1598

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table A3 Permitted/protected stimuli
Signal face arrangement No 8
Left turn signal characteristics Through signal characteristics
Number of signal heads: 1 Number of signal heads: 1
Vertical position: Overhead Vertical position: Overhead
Horizontal position: Between lanes Horizontal position: Over through lanes
Supplemental sign: None Special notes: Exclusive left turn lane
Full color Operations
YA
GB GB GA GB GB GB GB
Left Through Left Through Left Through
Stimulus No 35 Stimulus NO 36 Stimulus No 37
Full color Operations
RB R8 R8 R8 R8 RB
YA
GA
Left Through Left Through Left Through
Stimulus No 38 Stimulus No 39 Stimulus NO 40
Flashing operations
FRB FRB
FYB FYB
Left Through Left Through
Stimulus NO 41 Stimulus NO 42

 

 

 

 

 

1599

 

Table A3 (Cont'd).

 

Signal face arrangement No 8

 

Left turn signal characteristics
Number of signal heads: 1
Vertical position: Overhead
Horizontal position: Between lanes
Supplemental sign:

Left turn yield on green ball

Through signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Over through lanes
Special notes: Exclusive left turn lane

 

Full color Operations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

GB GB GA

 

 

 

 

 

GB GB

 

 

 

 

 

 

 

 

Left Through
Stimulus No 44

Left Through
Stimulus NO 45

 

Signal face arrangement No 9

 

Left turn signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Left turn lane
centerline

Supplemental Sign: None

Through signal characteristics

Number of signal heads: 2

Vertical position: Overhead

Horizontal position: Over through lanes
Special notes: Exclusive left turn lane

 

Full color operations

 

 

 

 

 

 

 

 

 

 

R8 R8

 

 

 

 

 

 

 

 

YA

YA

 

 

 

 

 

GA GB GB

 

 

 

 

 

 

GB GB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Left Through
Stimulus NO 47

Left Through
Stimulus NO 49

Left Through
Stimulus NO 48

 

Full color Operations

 

 

 

 

 

RB RB

 

 

 

 

 

 

 

 

 

GA

 

 

 

 

63 GB

 

 

 

 

 

 

 

 

Left Through
Stimulus NO 50

 

Left Through
Stimulus No 51

 

 

 

 

 

 

Table A3 (cont'd).

 

Signal face arrangement No 10

 

Left turn signal characteristics
Number of signal heads: 1

Vertical position: Overhead
Horizontal position: Between lanes
Sup lemental sign: None

 

Full color operations

Through signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Over through lanes

5 ial notes: Exclusive left turn lane

 

 

 

 

Left Through
Stimulus NO 52

Left Through Left

Stimulus NO 53

Stimulus No 54

Through Left Through

Stimulus NO 55

Left Through
Stimulus No 56

 

Signal face arrangement No 11

 

Left turn signal characteristics

Number of signal heads: 1

Vertical position: Pedestal

Horizontal position: Far median between lanes
Supplemental sign: None

Through signal characteristics

Number of signal heads: 2

Vertical position: Overhead

Horizontal position: Over through lanes
Special notes: Exclusive left turn lane

 

 

Full color operations

 

 

Left

Through Left

Stimulus No 57

Left Through
Stimulus No 58

 

Stimulus No 59

Through Left Through

Stimulus No 60

Left Through
Stimulus NO 61

 

 

 

 

 

 

2(91

 

Table A3 (Cont’d).

 

Signal face arrangement NO 12

 

Left turn signal characteristics
Number Of signal heads: 1

Vertical position: Overhead
Horizontal position: Between lanes
Supplemental sign: None

Through signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Over through lanes

 

Special notes: Exclusive left turn lane

 

Full color operations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

GB GB GB GB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

GA YA GA

 

 

 

 

Left Through Left
Stimulus No 63 Stimulus No 64

Through Left
Stimulus NO 65

YA

 

 

 

Through Left Through
Stimulus NO 66

 

Signal face arrangement No 13

 

Left turn signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Left-turn lane
centerline

Supplemental sign:

Left-turn must yield on flashing yellow--

only stimulus 70

Through signal characteristics

Number Of signal heads: 2

Vertical position: Overhead

Horizontal position: Over through lanes
Special notes: Exclusive left turn lane

 

 

Full color operations

 

 

 

 

 

 

 

 

 

 

 

 

FYB

 

FYB

 

 

 

 

 

GB GB

 

 

 

 

 

GB GB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

YA GA

 

 

 

 

Left Through Left
Stimulus No 67 Stimulus NO 68

Through Left
Stimulus NO 69

 

 

 

 

Through Left Through
Stimulus NO 70

 

 

 

 

 

 

2(92

 

Table A3 (Cont’d).

 

Signal face arrangement No 14

 

Left turn signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Left-turn lane
centerline

Supplemental sign: None

Through signal characteristics

Number of signal heads: 2

Vertical position: Overhead

Horizontal position: Over through lanes

Special notes: Exclusive left turn lane
Delaware DOT display

 

Full color operations

 

 

 

 

 

 

 

 

 

 

 

FRA

 

 

 

 

 

 

 

 

 

YA

 

 

 

 

 

 

GA GB

 

 

 

 

GB GB

 

 

 

 

 

 

 

 

 

 

 

 

 

Left Through
Stimulus NO 71

Left
Stimulus NO 72

 

Left Through
Stimulus No 73

Through

 

Full color operations

 

 

 

 

 

 

 

 

 

RB RB

 

 

 

 

YA YB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Left Through
Stimulus NO 74

Left
Stimulus NO 75

Through

 

Signal face arrangement No 15

 

Left turn signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Centerline of left-
turn lane

Supplemental sign: None

Through signal characteristics

Number of signal heads: 2

Vertical position: Overhead

Horizontal position: Over through lanes
Special notes: Exclusive left turn lane

 

Michigan display

 

Full color operations

 

 

 

 

 

 

 

FRB

RB RB RB

 

 

 

 

 

 

YB

 

 

 

 

 

 

GB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Left Through
Stimulus No 76

Left
Stimulus No 77

Left Through
Stimulus No 78

Through

 

Full color Operations

 

 

 

 

 

RB

 

 

 

 

YB

 

 

 

 

GA

 

 

 

 

 

GB

 

 

 

 

 

 

 

Left Through
Stimulus No 79

Left
Stimulus NO 80

Through

 

 

 

 

 

2(93

 

Table A3 (Cont'd).

 

Signal face arrangement No 16

 

Left turn signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Left-turn lane
centerline

Supplemental sign: None

 

Through signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Over through lane

Special notes: Shared left turn lane
Vancouver BC display

 

Full color operations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

GB GB
FFGA
Left Through

Stimulus NO 81

 

Signal face arrangement No 17

 

Left turn signal characteristics
Number Of signal heads: 1

Vertical position: Overhead
Horizontal position: Between lanes
Supplemental sign: Left turn signal

 

Through signal characteristics

Number of signal heads: 1

Vertical position: Overhead

Horizontal position: Over through lanes

Special notes: Exclusive left turn lane
Dallas TX display

 

Full color Operations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

RB RB
RB
YB GB
GB YA
GA GA
Left Through Left Through Left Through

Stimulus NO 82 Stimulus NO 83

Stimulus NO 84

 

 

 

 

APPENDIX B

APPENDIX B

Comprehension analysis results

 

Table Bl General comprehension analysis

 

Miles driven per year by subject age

 

ANALYSIS OF VARIANCE

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 1364906141 454968713.8 2.6185 .0525
WITHIN GROUPS 174 30232392631 173749382.9

TOTAL 177 31597298772

Miles driven per year by subject age and sex

 

ANALYSIS OF VARIANCE

Sum of Mean Sig
Source of Variation Squares DF Square F of F
Main Effects 2607596643 4 651899160.768 3.952 .004
AGE 1450877701 3 483625900.301 2.932 .035
SEX 1242690502 1 1242690501.66 7.534 .007
Z-Uay Interactions 947877278 3 315959092.728 1.915 .129
AGE SEX 947877278 3 315959092.728 1.915 .129
Explained 3555473921 7 507924845.894 3.079 .004
Residual 28041824851 170 164951910.887
Total 31597298772 177 178515812.271

191 cases were processed.
13 cases (6.8 pct) were missing.

Miles driven per year by subject age and sex

 

flit

... MULTIPLE CLASSIFICATION ANALYSIS

Grand Mean = 12080.93
Adjusted for

Unadjusted lndependents

Variable + Category N Dev'n Eta Dev’n Beta
AGE
16 TO 30 YEARS OLD 47 -418.05 ~653.73
31 10 45 YEARS OLD 48 1860.74 1805.56
46 TO 60 YEARS OLD 43 2722.56 3051.55
61 TO 75 YEARS OLD 40 -4668.43 -4678.96
.21 .21
SEX
MALE 102 2200.45 2287.55
FEMALE 76 -2953.23 -3070.13
.19 .20

 

 

 

2(94

 

 

 

2(95

 

Table B1 (cont'd).

 

Percent correct answers by miles driven per year

 

ANALYSIS OF VARIANCE

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 .6745 .2248 5.9903 .0007
WITHIN GROUPS 174 6.5309 .0375
TOTAL 177 7.2054

Percent serious error by miles driven per year

ANALYSIS OF VARIANCE

 

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 .0076 .0025 .8628 .4616
WITHIN GROUPS 174 .5132 .0029
TOTAL 177 .5208

Percent correct answers by subject age

 

ANALYSIS OF VARIANCE

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 1.1453 .3818 10.9486 .0000
WITHIN GROUPS 187 6.5204 .0349
TOTAL 190 7.6657

Percent serious errors by subject age

 

ANALYSIS OF VARIANCE

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 .0276 .0092 3.2611 .0227
WITHIN GROUPS 187 .5282 .0028
TOTAL 190 .5558

 

 

 

 

 

 

 

Table Bl (cont' d).

 

Percent correct answers by subject years of driving experience

 

ANALYSIS OF VARIANCE

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 .9370 .3123 8.7418 .0000
WITHIN GROUPS 176 6.2883 .0357
TOTAL 179 7.2253

Percent serious errors by subject years Of driving experience

 

ANALYSIS OF VARIANCE

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 .0266 .0089 3.0794 .0289
WITHIN GROUPS 176 .5067 .0029
TOTAL 179 .5333

Percent correct answers by subject sex

ANALYSIS OF VARIANCE

 

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 1 .4124 .4124 10.7472 .0012
WITHIN GROUPS 189 7.2532 .0384
TOTAL 190 7.6657

_

Percent serious errors by subject sex

 

ANALYSIS OF VARIANCE

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 1 .0031 .0031 1.0713 .3020
WITHIN GROUPS 189 .5527 .0029
TOTAL 190 .5558

 

 

 

 

2(97

 

Table B2 Inter-interval comprehension analysis

 

Serious error rate: One-way analysis Of variance by subject age

 

STANDARD STANDARD

 

GROUP COUNT MEAN DEVIATION ERROR MINIMUM MAXIMUM 95 PCT CONF INT FOR MEAN
Age 15-30 50 .0527 .0716 .0101 .0000 .3333 .0323 TO .0730
Age 31-45 51 .0569 .0640 .0090 .0000 .2273 .0389 TO .0749
Age 46-60 48 .0801 .0862 .0124 .0000 .4161 .0551 TO .1052
Age 60+ 42 .0943 .0783 .0121 .0000 .2914 .0699 TO .1187
Sum of Mean Sig

Source of Variation Squares DF Square F of F
Main Effects .053 3 .018 3.151 .026

AGE .053 3 .018 3.151 .026
Explained .053 3 .018 3.151 .026
Residual 1.057 187 .006
Total 1.111 190 .006

—

Serious error rate: Multivariate analysis of variance, repeated measures, all age groups

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Permitted stimuli .083 .150 186 .062 .105
Red stimuli .037 .100 186 .023 .052
Nighttime stimuli .126 .146 186 .105 .147
Change interval stimuli .010 .032 186 .005 .015
Source of Variation SS DF MS F Sig of F
AGE .26 3 .09 5.65 .001
STIMULUS MEANING 1.47 3 .49 38.31 .000
AGE BY MEANING .13 9 .01 1.11 .354

 

Serious error rate: Multivariate analysis of variance, repeated measures, older subjects

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Permitted stimuli .125 .172 42 .071 .179
Red stimuli .083 .142 42 .039 .127
Nighttime stimuli .154 .146 42 .109 .200
Change interval stimuli .014 .036 42 .003 .026
Source of Variation SS DF MS F Sig of F
STIMULUS MEANING .46 3 .15 9.73 .000

 

 

 

2(98

 

Table 82 (cont’ d).

 

Correct answer rate: Multivariate analysis of variance, repeated measures, all age groups

 

 

Mean Std. Dev. N 95 pct Conf.Interval

Permitted stimuli .558 .317 186 .513 .604
Protected stimuli .642 .319 186 .596 .688
Red stimuli .963 .100 186 .948 .977
Nighttime stimuli .446 .259 186 .408 .483
Change interval stimuli .740 .217 186 .709 .772
Source of Variation SS DF MS F Sig of F
AGE 5.43 3 1.81 13.37 .000

STIMULUS MEANING 29.04 4 7.26 180.82 .000
AGE BY MEANING 1.17 12 .10 2.42 .004

Correct answer rate: Multivariate analysis of variance, repeated measures, Older subjects

 

 

Mean Std. Dev. N 95 percent Conf.

Interval

Permitted stimuli .384 .300 42 .290 .478
Protected stimuli .488 .321 42 .388 .588
Red stimuli .917 .142 42 .873 .961
Nighttime stimuli .324 .206 42 .260 .388
Change stimuli .629 .274 42 .544 .715
Source Of Variation SS DF MS F Sig of F

STIMULUS MEANING 9.37 4 2.34 47.38 .000

 

 

 

 

 

 

 

 

Table 83 Intra-interval comprehension analysis

 

PERMITTED STIMULI

 

Serious error rate: One-way anal

ysis of variance by subject age

 

 

STANDARD STANDARD
GROUP COUNT MEAN DEVIATION ERROR MINIMUM MAXIMUM 95 PCT CONF INT FOR MEAN
Age 15-30 50 .0754 .2002 .0283 .0000 1.0000 .0185 TO .1323
Age 31'45 51 .0906 .1827 .0256 .0000 .8750 .0392 TO .1419
Age 46-60 47 .1120 .1851 .0270 .0000 .8000 .0577 TO .1664
Age 60+ 42 .1250 .1721 .0266 .0000 .7000 .0713 TO .1786
TOTAL 190 .0995 .1853 .0134 .0000 1.0000 .0730 TO .1260
ANALYSIS OF VARIANCE
SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 .0677 .0226 .6537 .5816
WITHIN GROUPS 186 6.4213 .0345
TOTAL 189 6.4890

Serious error rate: Multivariate analysis Of variance repeated measures all subjects

 

AGREETH Left:turn signal concurrence/discordance with through signal

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Concurrent .132 .240 187 .097 .167
Discordant .014 .064 187 .005 .023
Source of Variation SS DF MS F Sig of F
AGREETH 1.31 1 1.31 43.73 .000

Serious error rate: Multivariate analysis Of variance repeated measures all subjects

 

 

 

 

NOTHRU: Number of through signal heads
Mean Std. Dev. N 95 percent Conf. Interval
1 Through signal .104 .239 189 .069 .138
2 Through signals .098 .167 189 .074 .122
Source of Variation SS DF MS F Sig of F
NOTHRU .00 1 .00 .22 .643

 

 

 

 

21(9

 

Table BB (cont' d).

 

PERMITTED STIMULI

 

Serious error rate: Multivariate analysis of variance repeated measures all subjects

 

HPOSIT: Left-turn signal horizontal position

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Between lanes .128 .253 188 .092 .165
Straight-ahead .065 -135 188 .046 .085
Source of Variation SS DF MS F Sig of F
HPOSIT .37 1 .37 17.91 .000

Serious error rate: Multivariate analysis Of variance repeated measures all subjects

 

AGE, AGREETH: Left-turn signal concurrence/discordance with through signal

 

 

 

Concurrent
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .080 .207 49 .021 .140
AGE 31 TO 45 .137 .276 50 .058 .215
AGE 46 TO 60 .140 .222 46 .075 .206
AGE 61 TO 75 .178 .249 42 .100 .255
Discordant
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 IO 30 .005 .036 49 -.005 .015
AGE 31 TO 45 .015 .060 50 -.002 .032
AGE 46 TO 60 .018 .070 46 -.003 .039
AGE 61 TO 75 .018 .085 42 -.009 .044
Source of Variation SS OF MS F Sig of F
AGE .15 3 .05 1.53 .207
AGREETH 1.33 1 1.33 44.61 .000
AGE BY AGREETH .08 3 .03 .92 .431

 

 

r

 

 

 

 

 

2211

 

Table B3 (cont' d).

 

PERMITTED STIMULI

 

Serious error rate: Multivariate analysis of

variance repeated measures all subjects

 

AGE, NOTHRU: Number of through signal heads

 

1 Through signal

 

 

FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .090 .239 50 .022 .158
AGE 31 TO 45 .118 .267 50 .042 .194
AGE 46 TO 60 .105 .234 47 .036 .174
AGE 61 TO 75 .102 .217 42 .034 .169
2 Through signals
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .063 .181 50 .012 .115
AGE 31 TO 45 .074 .146 50 .033 .115
AGE 46 TO 60 .116 .162 47 .068 .163
AGE 61 TO 75 .147 .171 42 .093 .200
Source of Variation SS DF MS F Sig of F
AGE .12 3 .04 .55 .652
NOTHRU .00 1 .00 .08 .775
AGE BY NOTHRU .11 3 .04 2.40 .069

 

 

 

 

2212

 

Table B3 (Cont'd).
PERMITTED STIMULI

 

 

Serious error rate: Multivariate analysis of variance repeated measures all subjects

 

AGE, HPOSIT: Left-turn signal horizontal ppsition

 

Between lanes

 

FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .098 .246 50 .028 .168
AGE 31 TO 45 .117 .258 49 .043 .191
AGE 46 TO 60 .146 .252 47 .072 .220
AGE 61 TO 75 .158 .260 42 .077 .239
Straight-ahead

FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .046 .163 50 .000 .092
AGE 31 TO 45 .053 .115 49 .020 .086
AGE 46 TO 60 .069 .116 47 .035 .103
AGE 61 TO 75 .099 .135 42 .057 .141
Source of Variation SS DF MS F Sig Of F
AGE .17 3 .06 .93 .428
HPOSIT .37 1 ' .37 17.63 .000
AGE BY HPOSIT .01 3 .00 .13 .944

Serious error rate: Multivariate analysis of variance repeated measures older subjects

AGREETH: Left:turn signal concurrence/discordance with through signal

 

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Concurrent .178 .249 42 .100 .255
Discordant .018 .085 42 -.009 .044
Source Of Variation SS DF MS F Sig Of F
AGREETH .54 1 .54 15.23 .000

_

Serious error rate: Multivariate analysis of variance repeated measures older subjects

 

NOTHRU: Number of through signal heads

 

 

 

 

Mean Std. Dev. N 95 percent Conf. Interval
1 Through signal .102 .217 42 .034 .169
2 Through signals .147 .171 42 .093 .200
Source of Variation SS OF MS F Sig of F
NOTHRU .04 1 .04 2.56 .117
.IIIIIIII ‘TIIIIIIIIIIIIIIIIIIIIII

 

 

 

 

2213

 

Table B3 (Cont’d).
PERMITTED STIMULI

 

 

Serious error rate: Multivariate analysis of variance repeated measures Older subjects

 

HPOSIT: Left-turn signal horizontal position

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Between lanes .158 .260 42 .077 .239
Straight-ahead .099 .135 42 .057 .141
Source of Variation SS DF MS F Sig Of F
HPOSIT .07 1 .07 2.81 .102

-_—_——__l

Correct answer rate: One-way analysis of variance by subject age

 

STANDARD STANDARD

GROUP COUNT MEAN DEVIATION ERROR MINIMUM MAXIMUM 95 PCT CONF INT FOR MEAN
Age 15-30 50 .6573 .2878 - .0407 .0000 1.0000 .5755 TO .7391
Age 31-45 51 .6221 .3355 .0470 .0000 1.0000 .5278 TO .7165
Age 46-60 47 .4983 .2971 .0433 .0000 1.0000 .4111 TO .5856
Age 60+ 42 .3840 .3004 .0464 .0000 1.0000 .2904 TO .4776
TOTAL 190 .5481 .3218 .0233 .0000 1.0000 .5021 TO .5942

 

ANALYSIS OF VARIANCE

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 2.1234 .7078 7.5446 .0001
WITHIN GROUPS 186 17.4497 .0938
TOTAL 189 19.5731

_

Correct answer rate: Multivariate analysis of variance repeated measures all subjects

 

AGREETH: Left-turn signal concurrence/discordance with through signal

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Concurrent .566 .367 187 .513 .619
Discordant .545 .340 187 .496 .594
Source of Variation SS DF MS F Sig Of F
AGREETH .04 I .04 .66 .416

 

 

 

 

 

 

 

Table BB (Cont’d).

 

PERMITTED STIMULI

 

Correct answer rate: Multivariate analysis of variance repeated measures all subjects

 

NOTHRU: Number of through signal heads

 

 

Mean Std. Dev. N 95 percent Conf. Interval
1 Through signal .562 .383 189 .507 .617
2 Through signals .541 .315 189 .496 .586
Source of Variation SS of MS F Sig of F
NOTHRU .04 I .04 1.15 .286

 

Correct answer rate: Multivariate analysis of variance

repeated measures all subjects

 

HPOSIT: Left-turn signal horizontal position

 

Mean Std. Dev. N 95 percent Conf. Interval
Between lanes .563 .382 188 .508 .618
Straight:ahead .538 .320 188 .492 .584
Source of Variation SS OF MS F Sig of F
HPOSIT .06 1 .06 1.32 .252

 

Correct answer rate: Multivariate analysis of variance

repeated measures all subjects

 

AGE, AGREETH: Left-turn signal concurrence/discordance

with through signal

 

 

 

Concurrent
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .692 .341 49 594 790
AGE 31 TO 45 .672 .340 50 .575 768
AGE 46 T0 60 .519 .345 46 .416 621
AGE 61 TO 75 .346 .348 42 23B 455
Discordant
FACTO CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 To 30 .643 284 49 .561 .724
AGE 31 TO 45 .578 365 50 .475 682
AGE 46 TO 60 .495 332 46 .396 593
AGE 61 TO 75 448 352 42 .339 .558
AGE 4.04 3 1.35 7.95 .000
AGREETH .02 1 .02 41 524
AGE BY AGREETH .47 3 .16 2.60 053

 

 

 

 

Table B3 (Cont'd).

 

PERMITTED STIMULI

 

Correct answer rate: Multivariate analysis of variance repeated measures all subjects

 

AGE, NOTHRU: Number of through signal heads

 

1 Through signal

 

Correct answer rate: Multivariate analysis of variance repeated measures older subjects

FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .659 .365 50 .555 .762
AGE 31 TO 45 .668 .350 50 .568 .768
AGE 46 TO 60 .529 .381 47 .417 .641
AGE 61 TO 75 .359 .368 42 .244 .474
2 Through signals
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .657 .269 50 .581 .734
AGE 31 TO 45 .608 .338 50 .512 .704
AGE 46 TO 60 .472 .286 47 .388 .556
AGE 61 TO 75 .401 .302 42 .307 .495
Source of Variation SS DF MS F Sig of F
AGE 4.59 3 1.53 8.15 .000
NOTHRU .03 1 .03 .94 .333
AGE BY NOTHRU .16 3 .05 1.48 .221

 

AGREETH: Left-turn signal concurrence/discordance with through signal

 

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Concurrent .346 .348 42 .238 .455
Discordant .448 .352 42 .339 .558
Source of Variation SS DF MS F Sig of F
AGREETH .22 I .22 3.18 .082

—

 

 

 

2:16

 

Table 83 (Cont'd).

 

PERMITTED STIMULI

 

Correct answer rate: Multivariate analysis of variance repeated measures all subjects

 

AGE, HPOSIT: Left turn signal horizontal position

 

Between lanes

 

 

 

FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 T0 30 .675 .369 50 .570 .780
AGE 31 TO 45 .685 .338 49 .588 .783
AGE 46 TO 60 .505 .367 47 .398 .613
AGE 61 TO 75 .351 .365 42 .238 .465
Straight-ahead
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 T0 30 .643 .273 50 .565 .720
AGE 31 TO 45 .599 .336 49 .502 .696
AGE 46 TO 60 .471 .310 47 .380 .562
AGE 61 TO 75 .418 .316 42 .319 .516
Source of Variation 88 DF MS F Sig of F
AGE 4.67 3 1.56 8.56 .000
HPOSIT .04 1 .04 1.01 .316
AGE BY HPOSIT .27 3 .09 2.12 .100

 

 

 

 

 

 

 

Table B3 (Cont’d).

 

PERMITTED STIMULI

 

Correct answer rate: Multivariate analysis of variance repeated measures

older subjects

 

NOTHRU: Number of through signal heads

 

Mean Std. Dev. N 95 percent Conf. Interval
1 Through signal .359 .368 42 .244 .474
2 Through signals .401 .302 42 .307 .495
Source of Variation SS DF MS F Sig of F
NOTHRU .04 1 .04 .86 .360

 

Correct answer rate: Multivariate analysis of variance repeated measures

older subjects

 

HPOSIT: Left~turn signal horizontal position

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Between lanes .351 .365 42 .238 .465
Straightrahead .418 .316 42 .319 .516
Source of Variation SS DF MS F Sig of F
HPOSIT .09 1 .09 1.81 .186

 

 

 

 

 

 

2218

 

Table 83 (Cont'd).

 

PROTECTED STIMULI

 

Minor error rate: One-way analysis of variance by subject age

 

 

STANDARD STANDARD
GROUP COUNT MEAN DEVIATION ERROR MINIMUM MAXIMUM 95 PCT CONF INT FOR MEAN
Age 15-30 50 .2430 .2787 .0394 .0000 1.0000 .1638 TO .3222
Age 31-45 51 .3666 .3363 .0471 .0000 1.0000 .2720 TO .4612
Age 46-60 48 .3891 .3226 .0466 .0000 1.0000 .2955 TO .4828
Age 60+ 42 .5117 .3210 .0495 .0000 1.0000 .4117 TO .6117
TOTAL 191 .3718 .3264 .0236 .0000 1.0000 .3252 TO .4184
ANALYSIS OF VARIANCE
SUM OF MEAN F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 1.6673 .5558 5.5951 .0011
UITHIN GROUPS 187 18.5743 .0993
TOTAL 190 20.2415

Minor error rate: Multivariate analysis of variance repeated measures all subjects

 

SIGN: Left-turn supplemental sign

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Sign: "Left-turn signal" .311 .336 189 .263 .360
No sign present .378 .330 189 .331 .426
Source of Variation SS 0F MS F Sig of F

 

SIGN .42 1 .42 22.17 .000

Minor error rate: Multivariate analysis of variance repeated measures all subjects

 

COLORS: Left-turn and through signal lens configuration

 

 

 

Mean Std. Dev. N 95 percent Conf. Interval
GA, GA+GB Through: R8 .235 .355 188 .184 .286
GA+RB Through: R8 .438 .393 188 .382 .495
GA, GA+GB Through: GB .311 .395 188 .254 .368
GA+RB Through: GB .534 .335 188 .486 .582
Source of Variation SS DF MS F Sig of F
COLORS 9.94 3 3.31 69.72 .000

 

 

 

 

 

 

 

 

Table B3 (Cont’d).

 

PROTECTED STIMULI

 

Minor error rate: Multivariate analysis of variance repeated measures all subjects

 

AGE, SIGN: Left-turn '“nn'cm at ' sign

 

Sign: "Left-turn signal"
FACTOR CODE

 

 

Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 215 .307 50 .128 .302
AGE 31 To 45 295 .338 50 .199 .391
AGE 46 TO 60 316 .328 48 .221 .411
AGE 61 TO 75 443 .346 41 .334 .552
No sign present
FACTOR CODE Mean Std. Dev N 95 percent Conf. Interval
AGE 16 TO 30 250 281 50 .170 .330
AGE 31 T0 45 368 .337 50 .272 .464
AGE 46 TO 60 408 .338 48 .310 .506
AGE 61 T0 75 513 317 41 412 .613
Source of Variation SS DF MS F Sig of F
AGE 2.76 3 .92 4.81 .003
SIGN .43 I .43 22.16 .000
AGE BY SIGN .04 3 .01 .71 .545

 

 

 

 

 

 

Table 83 (Cont’d).

 

PROTECTED STIMULI

 

Minor error rate: Multivariate analysis of variance repeated measures all subjects

 

AGE, COLORS: Left-turn and through signal lens configuration

 

GA, GA*GB Through: RB
FACTOR CODE

 

Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .147 .305 50 .060 .234
AGE 31 T0 45 .212 .352 49 .111 .313
AGE 46 TO 60 .250 .351 48 .148 .352
AGE 61 TO 75 .353 .397 41 .227 .478
GAORB Through: RB
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .256 .353 50 .156 .356
AGE 31 To 45 .436 .401 49 .321 .551
AGE 46 TO 60 .503 .376 48 .394 .613
AGE 61 TD 75 .587 .375 41 .468 .705
Source of Variation SS DF MS F Sig of F
AGE 6.31 3 2.10 5.55 .001
COLORS 10.04 3 3.35 70.50 .000
AGE BY COLORS .47 9 .05 1.11 .352

 

Minor error rate: Multivariate analysis of variance repeated measures older subjects

 

SIGN: Left-[urn cnnnlnmpnfnl Sign

 

 

 

 

 

 

COLORS 3.01 3 1.00 17.93 .000

 

Mean Std. Dev. N 95 percent Conf. Interval
Sign: "Left-turn signal” .443 .346 41 .334 .552
No sign present .513 .317 41 .412 .613
Source of Variation SS DF MS F Sig of F
SIGN .10 1 .10 5.82 .021
Minor error rate: Multivariate analysis of variance repeated measures older subjects
COLORS: Left-turn and through signal lens configuration
Mean Std. Dev. N 95 percent Conf. Interval
GA, GA*GB Through: RE .353 .397 41 .227 .478
GA#RB Through: R8 .587 .375 41 .468 .705
GA, GA+GB Through: GB .426 .405 41 .298 .554
GA+R Through: GB .699 .302 41 .604 .795
Source of Variation SS DF MS F Sig of F

 

 

 

 

 

Table B3 (Cont'd).

 

RED STIMULI

 

Serious error rate: One-way analysis of variance by age

 

 

 

STANDARD STANDARD
GROUP COUNT MEAN DEVIATION ERROR MINIMUM MAXIMUM 95 PCT CONF INT FOR MEAN
Age 15-30 49 0136 .0432 .0062 .0000 .2222 .0012 TO .0260
Age 31-45 51 .0212 1074 0150 .0000 .7500 -.0090 TO .0514
Age 45-60 45 .0377 .0728 0109 .0000 .3333 .0158 TO .0595
Age 60+ 42 .0833 .1416 0219 .0000 .6250 .0392 TO 1275
TOTAL 187 .0371 .0998 0073 .0000 .7500 .0227 TO .0515
ANALYSIS OF VARIANCE
SUM OF EAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETHEEN GROUPS 3 .1297 .0432 4.5942 .0040
HITHIN GROUPS 183 1.7216 .0094
186 1.8513
Serious error rate: Multivariate analysis of variance repeated measures all subjects

 

HPOSIT: Left-turn signal horizontal position

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Straight-ahead .031 .224 191 -.001 .063
Between lanes .004 .174 191 -.021 .029
Source of Variation SS DF MS F Sig of F
HPOSIT .07 1 .07 6.44 .012

 

Serious error rate: Multivariate analysis of variance repeated measures

all subjects

 

NoTNRU: Number of through signal heads

 

 

Mean Std. Dev. N 95 percent Conf. Interval
1 Through signal .010 .244 191 -.024 .045
2 Through signals .014 .181 191 -.012 .040
Source of Variation SS DF MS F Sig of F
NOTHRU .00 1 .00 .07 .794

 

 

 

 

 

 

 

Table B3 (Cont'd).

 

RED STIMULI

 

Serious error rate: Multivariate analysis of variance repeated measures all subjects

 

 

 

 

 

 

 

 

SIGN: Left-turn ' ‘ ' sign
Mean Std. Dev. N 95 percent Conf. Interval
Sign: "Left turn signal" .025 .227 191 -.007 .058
No sign .008 .177 191 -.017 .034
Source of Variation SS DF MS F Sig of F
SIGN .03 1 03 2.41 .122
Serious error rate: Multivariate analysis of variance repeated measures all subjects
AGE, HPOSIT: Left-turn signal horizontal position
Straight-ahead
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 To 30 -.013 .223 50 - 077 050
AGE 31 To 45 .033 .153 51 - 010 076
AGE 46 To 60 -.007 .288 48 ~ 090 077
AGE 61 To 75 .127 .191 42 .068 186
Between lanes
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 To 30 -.013 146 50 -.055 .028
AGE 31 TO 45 .013 .073 51 -.008 .034
AGE 46 T0 60 -.037 .261 48 -.113 .038
AGE 61 TO 75 .061 .154 42 .013 .109
Source of Variation SS DF MS F Sig of F
AGE .74 3 .25 3.73 .012
HPOSIT .08 1 .08 7.24 .008
AGE BY HPOSIT .05 3 .02 1.56 .200

 

 

 

2223

 

Table 83 (Cont'd).
RED STIMULI

 

 

Serious error rate: Multivariate analysis of variance repeated measures all subjects

 

AGE, NOTHRU: Number of through signal heads

 

1 Through signal

 

 

FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 -.007 .158 50 -.052 .038
AGE 31 TO 45 .026 .147 51 -.015 .067
AGE 46 TO 60 -.021 .287 48 -.104 .062
AGE 61 TO 75 .048 .350 42 -.061 .157
2 Through signals
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 T0 30 -.007 .150 50 -.049 .036
AGE 31 TO 45 .020 .098 51 -.008 .047
AGE 46 TO 60 -.031 .263 48 -.108 .045
AGE 61 TO 75 .083 .158 42 .034 .133
Source of Variation SS DF MS F Sig of F
AGE .43 3 .14 1.92 .128
NOTHRU .00 1 .00 .12 .728
AGE BY NOTHRU .03 3 .01 .57 .636

—

Serious error rate: Multivariate analysis of variance repeated measures all subjects

 

AGE, SIGN: Left-turn supplemental sign

 

Sign: "Left turn signal"

 

 

FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 T0 30 -.013 .223 50 -.077 .050
AGE 31 TO 45 .033 .153 51 -.010 .076
AGE 46 T0 60 v.028 .274 48 -.107 .052
AGE 61 T0 75 .123 .224 42 .053 .193
No sign
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 -.013 .146 50 -.055 .028
AGE 31 TO 45 .016 .096 51 -.011 .043
AGE 46 TO 60 -.027 .270 48 -.105 .051
AGE 61 TO 75 .065 .141 42 .021 .109
Source of Variation SS DF MS F Sig of F
AGE .79 3 .26 3.82 .011
SIGN .03 1 .03 2.85 .093
AGE BY SIGN .05 3 .02 1.48 .222

 

 

 

 

   

 

2211

 

Table 83 (Cont’d).
RED STIMULI

 

 

Serious error rate: Multivariate analysis of variance repeated measures older subjects

 

HPOSIT: Left-turn signal horizontal position

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Straight-ahead .127 .191 42 .068 .186
Between lanes .061 .154 42 .013 .109
Source of Variation SS DF MS F Sig of F
HPOSIT .09 1 .09 4.96 .031

 
   

Serious error rate: Multivariate analysis of variance repeated measures older subjects

 

NOTHRU: Number of through signal heads

 

 

 

Mean Std. Dev. N 95 percent Conf. Interval
1 Through signal .048 .350 42 -.061 .157
2 Through signals .083 .158 42 .034 .133
Source of Variation SS DF MS F Sig of F
NOTHRU .03 1 .03 .44 .509

Serious error rate: Multivariate analysis of variance repeated measures older subjects

 

SIGN: Left-turn supplemental sign

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Sign: "Left turn signal" .123 .224 42 .053 .193
No sign .065 .141 42 .021 .109
Source of Variation SS 0F MS F Sig of F
SIGN .07 1 ‘ .07 3.61 .065

 

 

 

 

 

2225

 

Table 83 (Cont'd).
CHANGE INTERVAL STIMULI

 

 

Serious error rate: One-way analysis of variance by age

 

STANDARD STANDARD

GROUP COUNT MEAN DEVIATION ERROR MINIMUM MAXIMUM 95 PCT CONF INT FOR MEAN
Age 15-30 49 .0262 .0828 .0118 .0000 .3750 .0025 TO .0500
Age 31-45 50 .0200 .0813 .0115 .0000 .3750 -.0031 TO .0431
Age 46:60 47 .0137 .0479 .0070 .0000 .2500 -.0004 TO .0277
Age 60+ 42 .0304 .0721 .0111 .0000 .2500 .0079 TO .0528

 

ANALYSIS OF VARIANCE

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 .0072 .0024 .4574 .7124
WITHIN GROUPS 184 .9714 .0053
TOTAL 187 .9787

 

 

 

I__— ——_.l

 

 

 

 

2226

 

Table 83 (Cont'd).
CHANGE INTERVAL STIMULI

 

 

Correct answer rate: One-way analysis of variance by age

 

STANDARD STANDARD

GROUP COUNT MEAN DEVIATION ERROR MINIMUM MAXIMUM 95 PCT CONF INT FOR MEAN
Age 15-30 49 .8027 .1893 .0270 .0000 1.0000 .7483 TO .8571
Age 31-45 51 .7873 .1818 .0255 .0000 1.0000 .7361 TO .8384
Age 46-60 47 .7163 .2117 .0309 .0000 1.0000 .6541 TO .7785
Age 60* 42 .6293 .2739 .0423 .0000 1.0000 .5439 TO .7147
TOTAL 189 .7385 .2231 .0162 .0000 1.0000 .7065 TO .7705

 

ANALYSIS OF VARIANCE

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 .8471 .2824 6.1384 .0005
WITHIN GROUPS 185 8.5104 .0460
TOTAL 188 9.3576

Correct answer rate: Multivariate analysis of variance repeated measures all subjects

 

CHANGE: Change interval configuration

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Yellow ball or arrow .883 .254 182 .846 .920
Yellow arrow + Red ball .797 .274 182 .757 .837
Yellow arrow + Green ball .519 .397 182 .461 .577
Source of Variation SS DF MS F Sig of F
CHANGE 13.16 2 6.58 77.89 .000

Correct answer rate: Multivariate analysis of variance repeated measures all subjects

 

THRUCOL: Through signal indication

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Red ball .823 .270 186 .784 .862
Green ball .673 .255 186 .636 .710
Source of Variation SS DF MS F Sig of F
THRUCOL 2.09 1 2.09 43.72 .000

 

 

 

 

 

 

Table 83 (Cont’d).

 

CHANGE INTERVAL STIMULI

 

Correct answer rate: Multivariate analysis of variance

repeated measures

all subjects

 

AGEI GNANGE: Change interval configuration

 

Yellow ball or arrow

 

 

CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 To 30 .911 239 48 .842 .981
AGE 31 To 45 .920 239 48 .851 .989
AGE 46 TO 60 .924 189 46 .868 .980
AGE 61 To 75 .756 315 40 .656 .857
Yellow arrow + Red ball
FACTOR C Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 T0 30 .870 .217 48 .807 .933
AGE 31 T0 45 .829 .225 48 .764 .894
AGE 46 TO 60 .812 .244 46 .740 .885
AGE 61 TO 75 .654 .363 40 .538 .771
Yellow arrow + Green ball
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .547 .392 48 .433 661
AGE 31 10 45 .576 .383 48 .465 688
AGE 46 To 60 .433 .404 46 .313 553
AGE 61 To 75 .516 .411 40 .385 648
Source of Variation SS 0F MS F Sig of F
AGE 1.53 3 .51 4.17 .007
CHANGE 12.77 2 6.38 76.77 .000
AGE BY CHANGE .97 6 .16 1.94 .074

 

 

 

 

Table 83 (Cont'd).
CHANGE INTERVAL STIMULI

 

 

Correct answer rate: Multivariate analysis of variance repeated measures all subjects

 

AGE, THRUCOL: Through signal indication

 

 

 

Red ball
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .897 .217 49 .834 .959
AGE 31 TO 45 .873 .215 49 .811 .934
AGE 46 T0 60 .850 .219 46 .785 .916
AGE 61 TO 75 .647 .352 42 .537 .757
Green ball
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .718 .245 49 .648 .789
AGE 31 TO 45 .712 .228 49 .646 .777
AGE 46 T0 60 .635 .253 46 .560 .710
AGE 61 10 75 .616 .286 42 .527 .705
Source of Variation SS DF MS F Sig of F
AGE 1.68 3 .56 6.82 .000
THRUCOL 1.99 1 1.99 43.05 .000
AGE BY THRUCOL .42 3 .14 3.02 .031

Correct answer rate: Multivariate analysis of variance repeated measures older subjects

 

CHANGE: Change interval configuration

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Yellow ball or arrow .756 .315 40 .656 .857
Yellow arrow + Red ball .654 .363 40 .538 .771
Yellow arrow + Green ball .516 .411 40 .385 .648
Source of Variation SS DF MS F Sig of F
CHANGE 1.16 2 .58 5.36 .007

 

_

Correct answer rate: Multivariate analysis of variance repeated measures older subjects

 

THRUCOL: Through signal indication

 

 

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Red ball .647 .352 42 .537 .757
Green ball .616 .286 42 .527 .705
Source of Variation SS DF MS F Sig of F
THRUCOL .02 1 .02 .34 .562
III-IIIIIIIIIIIIIIIIIII IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII-IIIIIIIIIIIIIIIIIIII

 

 

22$)

 

Table 83 (Cont’d).
FLASHING INTERVAL STIMULI

 

 

Serious error rate: One-way analysis of variance by subject age

 

STANDARD STANDARD

GROUP COUNT MEAN DEVIATION ERROR MINIMUM MAXIMUM 95 PCT CONF INT FOR MEAN
Age 15-30 50 .0989 .1364 .0193 .0000 .4545 .0601 TO .1376
Age 31-45 51 .1074 .1543 .0216 .0000 .4545 .0640 TO .1508
Age 46-60 48 .1445 .1496 .0216 .0000 .4706 .1011 TO .1880
Age 60+ 42 .1544 .1459 .0225 .0000 .4545 .1089 TO .1999
TOTAL 191 .1248 .1474 .0107 .0000 .4706 .1038 TO .1459

 

ANALYSIS OF VARIANCE

SUM OF MEAN F F
SOURCE D.F. SQUARES SQUARES RATIO PROB.
BETWEEN GROUPS 3 .1046 .0349 1.6202 .1862
WITMIN GROUPS 187 4.0260 .0215
TOTAL 190 4.1306

Serious error rate: Multivariate analysis of variance repeated measures all subjects

 

FLASH: Left°turn driver action during flashing signal operation

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Stop for both directions .272 .332 188 .224 .319
Stop for opposing direction .076 .164 188 .053 .100
Permitted .010 .065 188 .001 .019
Source of Variation SS DF MS F Sig of F
FLASH 6.96 2 3.48 83.74 .000

Serious error rate: Multivariate analysis of variance repeated measures all subjects

 

LFTFLSH: Left-turn signal indication during flashing signal operation

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Flash Red Ball .206 .242 184 .171 .241
Flash Yellow Ball .004 .034 184 -.001 .009
Dark left-turn signal .124 .198 184 .095 .153
Source of Variation SS DF MS F Sig of F
LFTFLSH 3.79 2 i 1.89 87.65 .000

 

 

—

 

Table 83 (Cont'd).

 

FLASHING INTERVAL STIMULI

 

Serious error rate: Multivariate analysis of variance repeated measures all subjects

 

THRUCOL: Through signal indication

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Flash Red Ball .270 .331 189 .223 .318
Flash Yellow Ball .185 .227 189 .153 .218
Source of Variation SS DF MS F Sig of F
THRUCOL .68 .68 109.70 .000

 

Serious error rate: Multivariate analysis of variance repeated measures

all subjects

 

Stop for both directions
FACTOR

AGE, FLASH: Left-turn driver action during flashing signal operation

 

 

Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 T0 30 .215 .318 49 .124 .307
AGE 31 TO 45 .241 .342 51 144 .337
AGE 46 TO 60 .314 .322 47 .219 .408
AGE 61 T0 75 .330 .341 41 222 437
Stop for opposing direction
FACTOR CODE Mean Std Dev N 95 percent Conf. Interval
AGE 16 TO 30 .073 171 49 .024 .122
AGE 31 TO 45 .060 144 51 .020 .101
AGE 46 TO 60 .064 142 47 .022 .106
AGE 61 TO 75 114 198 41 .051 .176
Permitted
FACTOR CODE Mean Std Dev N 95 percent Conf. Interval
AGE 16 TO 30 .005 024 49 -.002 .012
AGE 31 TO 45 .000 000 51 .000 .000
AGE 46 TO 60 .022 103 47 -.009 .052
AGE 61 T0 75 .015 080 41 -.010 .040
Source of Variation SS DF MS F Sig of F
AGE .29 3 .10 1.67 .175
FLASH 7.05 2 3.52 84.67 .000
AGE BY FLASH .23 6 .04 .91 .485

 

 

 

2131

 

Table 83 (Cont’d).
FLASHING INTERVAL STIMULI

 

 

Serious error rate: Multivariate analysis of variance repeated measures all subjects

 

AGE, LFTFLSH: Left-turn signal indication during flashing signal operation

 

Flash Red Ball

 

 

FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 T0 30 .176 .239 47 .106 .247
AGE 31 TO 45 .187 .254 50 .115 .259
AGE 46 TO 60 .206 .223 45 .139 .273
AGE 61 TO 75 .262 .249 42 .185 .340
Flash Yellow Ball
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 T0 30 .000 .000 47 .000 .000
AGE 31 TO 45 .000 .000 50 .000 .000
AGE 46 TO 60 .017 .067 45 -.003 .038
AGE 61 TO 75 .000 .000 42 .000 .000
Dark left-turn signal
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 T0 30 .089 .177 47 .037 .141
AGE 31 TO 45 .108 .200 50 .051 .165
AGE 46 TO 60 .185 .226 45 .117 .253
AGE 61 TO 75 .117 .179 42 .061 .173
Source of Variation SS DF MS F Sig of F
AGE .21 3 .07 1.25 .292
LFTFLSH 3.84 2 1.92 89.90 .000
AGE BY LFTFLSH .23 6 .04 1.80 .097

 

 

 

 

 

 

 

 

 

 

 

 

Table 83 (Cont 'd) .
FLASHING INTERVAL STIMULI
Serious error rate: Multivariate analysis of variance repeated measures all subjects
AGE, THRUCOL: Through signal indication
Flash Red Ball
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .215 .318 49 .124 .307
AGE 31 TO 45 .241 .342 51 .144 .337
AGE 46 T0 60 .314 .322 47 .219 .408
AGE 61 TO 75 .322 .341 42 .215 .428
Flash Yellow Ball
FACTOR ODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 To 30 .144 .210 49 .083 .204
AGE 31 T0 45 .165 .238 51 .098 .232
AGE 46 To 60 .217 .223 47 .152 .283
AGE 61 To 75 .223 .233 42 .150 .295
Source of Variation SS DF MS F Sig of F
AGE .59 3 .20 1.28 .283
THRUCOL .69 1 .69 110.40 .000
AGE BY THRUCOL .01 3 .0 .70 .551

 

 

 

 

 

Table 83 (Cont'd).
FLASHING INTERVAL STIMULI

 

 

Serious error rate: Multivariate analysis of variance repeated measures older subjects

 

FLASH: Left-turn driver action during flashing signal operation

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Stop for both directions .330 .341 41 .222 .437
Stop for opposing direction .114 .198 41 .051 .176
Permitted .015 .080 41 -.010 .040
Source of Variation SS DF MS F Sig of F
FLASH 2.12 2 1.06 19.97 .000

Serious error rate: Multivariate analysis of variance repeated measures older subjects

 

LFTFLSH: Left-turn signal indication during flasing signal operation

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Flash Red Ball .262 .249 42 .185 .340
Flash Yellow Ball .000 .000 42 .000 .000
Dark left-turn signal .117 .179 42 .061 .173
Source of Variation SS DF MS F Sig of F
LFTFLSH 1.45 2 .72 36.07 .000

_

Serious error rate: Multivariate analysis of variance repeated measures older subjects

 

THRUCOL: Through signal indication

 

Mean Std. Dev. N 95 percent Conf. Interval

 

Flash Red Ball - .322 .341 42 .215 .428
Flash Yellow Ball .223 .233 42 .150 .295
Source of Variation SS DF MS F Sig of F
THRUCOL .21 1 .21 29.26 .000

 

 

 

Table 83(Cont’d).

 

FLASHING INTERVAL STIMULI

 

Correct answer rate: One-way analysis of variance by age

 

STANDARD STANDARD
GROUP COUNT MEAN DEVIATION ERROR MINIMUM

 

 

MAXIMUM 95 PCT CONF INT FOR MEAN

Age 15-30 50 .5517 .2546 .0360 .0000 1.0000 .4793 TO .6241
Age 31-45 51 .5022 .2746 .0385 .0000 1.0000 .4250 TO .5795
Age 46-60 48 .3779 .2416 .0349 .0000 .9524 .3078 TO .4481
Age 60* 42 .3239 .2056 .0317 .0000 .8182 .2599 TO .3880
TOTAL 191 .4447 .2615 .0189 .0000 1.0000 .4074 TO .4821

ANALYSIS OF VARIANCE

SUM OF MEAN F F

SOURCE D.F. SQUARES SQUARES RATIO PROB.

BETWEEN GROUPS 3 1.5680 .5227 8.5555 .0000
WITHIN GROUPS 187 11.4240 .0611
TOTAL 190 12.9920

 

Correct answer rate: Multivariate analysis of variance repeated measures all subjects

 

FLASH: Left-turn driver action during signal flahsing operation

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Stop for both directions .577 .389 188 .521 .633
Stop for opposing direction .369 .374 188 .315 .423
Permitted .338 .373 188 .284 .392
Source of Variation SS DF MS F Sig of F
FLASH 6.38 2 3.19 26.16 .000

 

Correct answer rate: Multivariate analysis of variance repeated measures all subjects

 

LFTFLSH: Left-turn signal indication during flashing signal operations

 

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Flash red bell .514 .277 184 .473 .554
Flash yellow ball .368 .415 184 .308 .428
Dark left-turn signal .406 .333 184 .358 .455
Source of Variation SS DF MS F Sis Of F

LFTFLSH 2.10 2 1.05 15.14 .000

 

 

 

 

Table 83 (Cont'd).

 

FLASHING INTERVAL STIMULI

 

Correct answer rate: Multivariate analysis of variance repeated measures all subjects

 

THRUCOL: Through signal

.J.

Mean Std. Dev.

N 95 percent Conf. Interval

 

 

 

 

 

 

Flash Red Ball .580 .390 189 .524 .635
Flash Yellow Ball .343 .325 189 .296 .389
Source of Variation SS DF MS F Sig of F
THRUCOL 5.30 1 5.30 43.96 .000
Correct answer rate: Multivariate analysis of variance repeated measures all subjects
AGE, FLASH: Left-turn driver action during signal flashing operation
Stop for both directions
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 To 30 .698 .366 49 .593 .803
AGE 31 TO 45 .614 .405 51 .500 .728
AGE 46 TO 60 .537 .374 47 .427 .647
AGE 61 TO 75 .434 .374 41 316 .552
Stop for opposing direction
CTO CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .350 .377 49 .242 .458
AGE 31 To 45 .436 .410 51 .321 .552
AGE 46 TO 60 .314 .352 47 .211 .417
AGE 61 TO 75 .370 348 41 .260 .480
Permitted
FACTOR CODE Mean Std. Dev N 95 percent Conf. Interval
AGE 16 T0 30 .458 .409 49 .341 .575
AGE 31 To 45 .420 .391 51 310 .530
AGE 46 10 60 .266 .326 47 .171 .362
AGE 61 T0 75 .174 .277 41 .087 .262
Source of Variation $5 or MS r 519 of F
AGE 3.12 3 1.04 6.01 .001
FLASH 6.27 2 3.14 26.11 .000
AGE BY FLASH 1.37 6 .23 1.91 .079

 

 

 

 

 

Table 83 (Cont’d).

FLASHING INTERVAL STIMULI

 

Correct answer rate: Multivariate analysis of variance repeated measures all

subjects

 

AGE, LFTFLSH: Left-turn signal indication during flashing signal operation

 

Flash Red Ball
FACTOR

 

 

CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .576 .233 47 .508 .645
AGE 31 TD 45 .566 .289 50 .484 648
AGE 46 TO 60 .476 .286 45 .390 .562
AGE 61 To 75 .421 .275 42 .336 .507
Flash yellow ball
FACTO CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 T0 30 .512 .442 47 .382 .642
AGE 31 To 45 .468 .434 50 .345 .591
AGE 46 TO 60 .285 .370 45 .174 .396
AGE 61 To 75 .177 .311 42 .081 .274
Dark left-turn signal
FACTOR CODE Mean Std. Dev N 95 percent Conf. Interval
AGE 16 TO 30 .565 341 47 .465 666
AGE 31 TO 45 .438 .337 50 .342 534
AGE 46 T0 60 .317 .288 45 .230 404
AGE 61 To 75 .286 .292 42 .195 .377
Source of Variation SS DF MS F Sig of F
AGE 5.59 3 1.86 9.60 .000
LFTFLSH 2.19 2 1.09 15.94 .000
AGE BY LFTFLSH .66 6 .11 1.61 .142

 

 

 

 

2137

 

Table 83 (Cont'd).
FLASHING INTERVAL STIMULI

 

 

Correct answer rate: Multivariate analysis of variance repeated measures all subjects

 

AGE, THRUCOL: Through signal indication

 

Flash Red Ball

 

FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 TO 30 .698 .366 49 .593 .803
AGE 31 TO 45 .614 .405 51 .500 .728
AGE 46 T0 60 .537 .374 47 .427 .647
AGE 61 T0 75 .447 .379 42 .329 .565
Flash Yellow Ball
FACTOR CODE Mean Std. Dev. N 95 percent Conf. Interval
AGE 16 T0 30 .427 .348 49 .327 .527
AGE 31 TO 45 .416 .343 51 .320 .513
AGE 46 TO 60 .279 .296 47 .192 .366
AGE 61 T0 75 .227 .259 42 .146 .307
Source of Variation SS DF - MS F Sig of F
AGE 2.86 3 .95 7.72 .000
THRUCOL 5.27 1 5.27 43.20 .000
AGE BY THRUCOL .08 3 .03 .23 .874

Correct answer rate: Multivariate analysis of variance repeated measures older subjects

FLASH: Left-turn driver action during flashing signal operation

 

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Stop for both directions .434 .374 41 .316 .552
Stop for opposing direction .370 .348 41 .260 .480
Permitted .174 .277 41 .087 .262
Source of Variation SS DF MS , F Sig of F
FLASH 1.50 2 .75 6.94 .002

Correct answer rate: Multivariate analysis of variance repeated measures older subjects

 

LFTFLSH: Left-turn signal indication during flashing signal operation

 

 

Mean Std. Dev. N 95 percent Conf. Interval
Flash Red Ball .421 .275 42 .336 .507
Flash yellow ball .177 .311 42 .081 .274
Dark left-turn signal .286 .292 42 .195 .377
Source of Variation SS DF MS F Sig of F
LFTFLSH 1.26 2 .63 9.51 .000

 

 

_

 

 

 

 

 

 

Table 83 (Cont’d).

 

FLASHING INTERVAL STIMULI

 

Correct answer rate: Multivariate analysis of variance repeated measures older subjects

 

THRUCOL: Through signal indication

 

 

Mean Std. Dev. N 95 percent Conf. Interval
F lash Red Ball .447 .379 42 .329 .565
F lash Yellow Ball .227 .259 42 .146 .307
Source of Variation SS DF MS F Sig of F
THRUCOL 1.02 1 1.02 8.46 .006

 

 

 

 

 

 

 

 

LI ST OF REFERENCES

LIST OF REFERENCES

2L - "A Time Progression Viewpoint of Accidents of Elderly
Drivers", N. Stamatiadis, B. C. Lin, Transportation
Research Board preprint paper No 920633, TRB
Washington, D. C., 1992.

:2 - "Older Driver Resource Directory", Transportation
Research Circular No 385, January 1992.

:3 - "Safety Issues Concerning Aged Drivers Response to
Traffic Control Devices". T.R. Circular. TRB Annual
Meeting 1988.

‘1 - "Accident Characteristics of Elderly Drivers at
Intersections", N.J. Garber, R. Srinivasan,
Transportation Research Board preprint paper No
910609, TRB Washington, D.C., 1991.

55 - Accidents Acts, 1976. National Safety Council,
Chicago, Ill., 1976.

£5 - Agent, Kenneth R., "Development of Warrants for Left-
Turn Lanes", Research Report 526, Kentucky Bureau of
Highways, Lexington, Kentucky, 1979.

'7 - Allen, T.M. "Nighttime Legibility Distances of
Highway Signs," HRB Bulletin, No. 191, 1958, pp. 33-

53- Ashton S.J. and G.M. Mackay. Benefits from Changes in
Vehicle Exterior Design. In Pedestrian Impact Injury
and Assessment, Society of Automotive Engineers,
Detroit Michigan, March 1983, pp 225-264.

9" Awadallah F.I., "Guidelines for the Adequacy of
Traffic Signs Retroreflectivity", ITE Compendium of
Technical Papers, 1991.

1C)- Bailey, I.L. and J.E. Sheedy. "Vision Screening for
Driver Licensure, "TRB Special Report 218, Vol 2,
1988, pp 294-324.

11“ Bailey 8.8., S.A. Jones, R.J. Stout, J.E. Bailey, S.
Kass, B.B. Morgan Jr., "Issues on Elderly
Pedestrians", 1991.

112 .

' Ball, K., Beard, 8., Roenker, D., Miller, R., &

Griggs, D., "Age and visual search: Expanding the

useful field of view", Journal of the Optical
Society of America A, 5, 2210-2219, 1988.

239

13.

15.

l6.

l7.

18.

20.

21.

22.

23.

24.

240

Ball, K., Owsley, C. & Beard, 8., "Clinical visual
perimetry underestimates peripheral field problems in
older adults", Clinical Vision Sciences, 5, 113-125,
1990.

Barrett G.V., W. Mihal, P.E. Panek, H.L. Sterns, and
R. A. Alexander. Information Processing Skills
Predictive of Accident Involvement for Younger and
Older Commercial Drivers. Industrial Gerontology,
Vol. 4, 1977, pp 173—182.

Bishu R.R., M. Tarawneh, P.T. McCoy, and B. Foster,
"A Predictive Model for Elderly Drivers",
Transportation Research Board preprint paper No
920712, TRB Washington, D.C., 1992.

Braune R. and C.D. Wickens. The Functional Age
Profile: An Objective Decision Criterion for the
Assessment of Pilot Performance Capacities and
Capabilities. Human Factors, Vol. 27, No. 6, 1985,
pp 681-693.

Brindle R.E., Some Aspects of Town Planning and Road
Safety. Presented at New Zealand Roadway Symposium,
Wellington, 1983.

Bull J. P. and G. M. Mackay. Some Characteristics of
Collisions, the Population of Car Occupant Casualties
and Their Relevance to Performance Testing. In
Proceedings of the Fourth International Conference on
Impact Trauma, International Research Committee on
the Biokinetics of Impacts, Lyons, France, September
1978, pp.l3—26.

Burden, D., "The older pedestrian: Special Problems
and Needs", unpublished data, Talahassee, Florida,
Florida Department of Transportation, 1991.

Burg A. Vision and Driving: A Report on Research.
Human Factors, Vol. 13, No. l, 1971, pp.79—87.

Burg A., Vision Test Scores and Driving Record:
Additional Findings. Report 68—27. Department of
Engineering, University of California, Los Angeles,
1968.

Burg A., The Relationship Between Vision Test Scores
and Driving Record: General Findings. Report 67-24.
Department of Engineering, University of California,
Los Angeles, 1967.

Burg, A. Lateral visual field as related to age and
sex. Journal of Applied Psychology, 52, 1968, pp.
10—15.

Cerrelli, E. Older Drivers, The Age Factor in
Traffic Safety. Report DOT-HS-807-402. U.S.
Department of Transportation, 1989.

 

 

241

25. Cottrell B.H., Jr. Guidelines for
Protected/Permissive Left-Turn Signal Phasing. In
transportation Research Record 1069, TRB, National
Research Council, Washington, D.C., 1986, pp. 54-61.

26. Curry J.D. The Mechanical Consequences of Variation
in the Mineral Content of Bone. Journal of
Biomechanics, Vol 2, 1969, pp. 1-11.

27. Davis G.A., K. Koutsoukos, "A Statistical Method for
Identifying Locations of High Crash Risk to Older
Drivers", Transportation Research Board preprint
paper No 920267, TRB Washington, D.C., 1992.

28. Davison P.A. Inter-relationship Between British
Drivers' Visual Abilities, Age and Road Accident
Histories. Ophthalmic and Physiological Optics, Vol.
5, 1985, pp 195-204.

 

29. Decina L.E., L.K. Staplin, A. Spiegel, "Correcting
Unaware Vision Impaired Drivers", final report
prepared by Ketron Inc. for PennDOT, Oct 1990.

30. "Effects of Supplemental Interchange Signing on Driver
Control Behavior", Technical Report, FHWA, Washington
D.C., March 1986.

31. Evans L. Fatal and Severe Crash Involvement Versus
Driver
32. Age and Sex. GMR-5721. General Motors Research

Laboratories, Warren, Michigan, 1987.

33. Evans L. Young Driver Involvement in Severe Car
Crashes. Report GMR-5835. General Motors Research
Laboratory, Warren, Michigan, 1987.

34. Evans L., "Older driver risks--to themselves and to
other road users", Transportation Research Board
preprint paper No 910490, TRB Washington, D.C., 1991.

35. Fatal Accident Reporting System: 1984. NHTSA,
U.S.Department of Transportation, March 1985.

36. "The Federal Highway Administration’s Action Plan for
Older Persons", U.S. DOT, February 1989.

37. Fishman, G.A., Anderson, R.J., Simson, L., & Hague,
A. (1981). Driving performance of retinitis
pigmentosa patients. British Journal of
Opthalmoloqv. 65 122-126.

38. Forbes, T.W., K.Moscowitz, G.Morgan, and D.W.
Loutzenheiser. "A Comparison of Lower Case and
Capital Letters for Highway Signs," TAB Proceedings,
Vol. 30, 1950, pp. 355-373.

39. Frisbie T., "Keeping Over-The-Hill Drivers in the
Game", Traffic Safety, January/February, 1991 pp 10-
13.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

SO.

51.

52.

242

Grubb, M., "Driver behavior at intersections: An
Analysis of Accident Related Variables", Proceedings

of the Human Factors Society. Santa Monica, CA:
Human Factors Society, 1986.

Guidelines for Major Street Design - A Recommended
Practice. Institute of Transportation Engineers,
Washington, D.C., 1984.

Harrington D. and R.McBride. Traffic Violations by
Type, Age, Sex, and Marital Status. Accident Analysis
and Prevention, Vol. 2, 1970, pp. 67-69.

Hawkins H.E., "A Comparison of Leading and Lagging
Greens in Traffic Signal Sequence," 1963 Proceedings,
Institute of Traffic Engineers, 33rd Annual Meeting,
Toronto, Ontario, Canada, pp. 238—242.

Henderson R.L., A. Burg, and F.A. Brazelton,
"Development of an Integrated Vision Testing Device:
Phase I", Final Report, Systems Development
Corporation Santa Monica, California, 1971.

Henderson, R. & Burg, A., "Vision and Audition in
Driving", Technical Report No. TM(L)—5297/000/00,
Washington D.C.: Department of Transportation, 1974.

Highway Statistics 1984. FHWA, U.S. Department of
Transportation, October 1985.

Hills B. L. and A. Burg. A Reanalysis of California
Driver Vision Data: General Findings. U.K. Transport
and Road Research Laboratory, Crowthorne, Berkshire,
England, 1977.H. W. Hofstetter. Visual Acuity and
Highway Accident. Journal of the American Optometric
Association, Vol. 47, 1976, pp. 887—893.

Hills, B.L., & Burg, A., "A Re—analysis of
California driver vision data: General findings",
Report 768. Transport and Road Research Laboratory,
Crowthorne, England, 1977.

Hostetter, R., McGee, H., Crowley, K., Sequin, E.,
and Dauber, G. (1986). Improved Perception—Reaction

Time Information for Intersection sight Distance.
FHWA/ R087/015. Washington, D.C.: Federal Highway

Administration.

Hummer J.E., R.E. Montgomery, K.C. Sinha, "Guidelines
for the Use of Leading and Lagging Left-Turn Signal
Phasing", presentation TAB, January 1991.

Hunter-Zaworski, K., "The Performance of Drivers
with Physical Limitations at T-Intersections",
Covallis, OR: Doctoral dissertation, Department of
Civil Engineering, Oregon State University, 1987.

IIHS Facts. Insurance Institute for Highway Safety,
Arlington VA., 1988.

 

S3.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

243

"Impact of Aging Driver Population and Impaired

Drivers on Geometric Design", Transportation Research
Circular No 317, June 1987.

"Implementation of Periodic Vision Screening to a

Statewide Program", Task 5 Report, prepared by Ketron
for PennDOT, January, 1990.

Johnson C.A. and J.L.Keltner. Incidence of Visual
Field Loss in 20,000 Eyes and Its Relationship to
Driving Performance. Archives of Opthalmology, Vol.
101, 1983, pp. 371-375.

Kausler D.H. Experimental Psychology and Human
Aging. John Wiley, New York, 1982.

Khavanin N.R. and R.N. Schwab, "Traffic Sign
Legibility and Conspicuity for the Older Drivers",
ITE Compendium of Technical Papers, 1991.

Kikuchi 8., J.R. Riegner, "A Methodology to Analyze
the Driver Decision Environment at a Signal Change
Interval: Application of Fuzzy Set Theory",
presentation at the 70th Annual Meeting of the TAB,
Dec 1990.

Kline, D.W., F.8chieber, L.A. Abusamara, and A.C.
Coyne. "Age, the Eye and the Visual Channels:
Contrast Sensitivity and Response Speed," Journal of
Gerontology, Vol. 38, No. 2, 1983, pp. 211-216.

Koltonow P. 6. Improving Safety and Mobility for
Older People. TR News, September-October 1985, pp
20-23.

Korteling, J., "Perception-response speed and driving
capabilities of brain-damaged and older drivers",

Human factors, 32, 95-108, 1990.

Leibowitz, H. & Appelle, 8., "The effect of a central
task on luminance thresholds for peripherally
presented stimuli", Human Factors, 11, 387-392,
1969.

Lerner N.D. and D.J. Ratte, "Problems in Freeway Use
as Seen By Older Drivers", Transportation Research
Board preprint paper No 910152, TAB Washington, D.C.,
1991.

Lerner N.D., "Older Driver Perception-Reaction Time
and Sight Distance Design Criteria", ITE Compendium
of Technical Papers, 1991.

Lerner N.D., "Some Observations on Older Drivers",
TAB, June 1988.

Lerner, N., Ratte, D., Huey, R., McGee, H., and
Hussain, 8., Literature Review: Older Driver
Perception-Reaction Time. Working paper under
contract DTFH 61-90-C—00038. Washington, D.C.:
Federal Highway Administration, 1990.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

0

244

Lin F.B., "Left-Turn Signal Phasing For Full-Actuated
Signal Control", Transportation Research Board
preprint paper No 910463, TAB Washington, D.C., 1991.

Lundgren-Linquist, B.; Aniansson, A.; & Rundgren, A.
Functional studied in 79-year-olds. III. Walking
performance and climbing ability. Scandinavian
Journal of Rehabilitation Medicine, 15, p. 125-131.

Mace D.J., R. Gable, "Model Highway Visibility
Minimum Required Visibility", Transportation Research
Board reprint paper No 920709, TAB Washington, D.C.,
1992.

Maleck T. and Hummer J. Driver Age and Highway
Safety. In Transportation Research Record 1059, TAB,
National Research Council, Washington, 1986, pp.6-12.

 

McCoy P.T., "Strategies for Improving the Safety of
Elderly Drivers", ITE Compendium of Technical Papers,
1991.

McGee, H., Hooper, K., Jughes, W., and Benson, W.
(1983). Highway Design and Operational Standards
Affected by Driver Characteristics. FHWA/RD-83/015.
Washington, D.C.: Federal Highway Administration.

McKelvey F.x. and Nikiforos Stamatiadis, "A Study of
the Relationship Between Older Drivers and highway
Safety in the State of Michigan", Final Report, CEE
Department, MSU E. Lansing, Michigan.

McKelvey F.X. and Nikiforos Stamatiadis. Accidents
and the older driver at signalized and non-signalized
intersections. Division of Engineering Research,
College of Engineering, Michigan State University,
East Lansing Michigan.

McKelvey F.X., N. Stamatiadis, "Highway Accident
Patterns in Michigan Related to Older Drivers",
Transportation Research Board reprint paper 88-0374,
Washington D.C. 1989.

McKelvey F.X., T.L. Maleck, N. Stamatiadis, P.
Stamatiadis, "Relationship Between Driver Age and
Highway Accidents", MSU, MOHSP, NHTSA, July 1987.

Moore R.L., I.P. Sedgley and B.E. Sabey. Ages of car
drivers involved in accidents, with special reference
to junctions. Transport and Road Research
Laboratory, Department of the Environment, Department
of Transport, Supplementary Report 718 Crowthorne,
Berkshire, 1982.

National Highway Traffic Safety Administration.
(1989a). Older Drivers. The Age Factor in Traffic
Safety. (Report Number DOT HS 807 402) Washington.
D.C.: U.S. Department of Transportation.

 

79. "The National Highway Traffic Safety Administration’s
Traffic Safety Plan for Older Persons", U.S. DOT,
NHTSA, Sep. 1988.

80. Nationwide Personal Transportation Study 1983-84.
FHWA, U.S. Department of Transportation, 1983.

81. "New Technology For Older Drivers", Traffic Safety,
January/February, 1991 p 5.

82. N9 J.C.N., E. Hauer, and J.Lovel. Are Longer Signal
Cycles Safer? In Proceedings, 12th Annual Meeting of
the Institute of Transportation Engineers, District
7, Canada, Hamilton Ontario, Canada June 1987.

83. NHTSA Conference on Research and Development Needed
to Improve Safety and Mobility of Older Drivers.
Washington, D.C.: National Highway Traffic Safety
Administration, 1990.

84. North R. V. The Relationship Between the Extent of
Visual Field and Driving Performance - A Review.
Ophthalmic and Physiological Optics, Vol.5, 1985, pp
205-210.

85. "Older Driver Pilot Program", Federal Highway
Administration, Office of Highway Safety, Report to
Congress Apr87-Apr90, Washington D. C., 990.

86. "Older drivers, The Age Factors in Traffic Safety",
NHTSA, February 1989.

87. Olson, P., Cleveland, 0., Fancher, P., Kostyniuk,L.,
and Schneider, L., "Parameters Affecting Stopping
Sight Distancez, NCHRP Report 270. Washington, D.C.:
Transportation Research Board, 1984.

88. "Operation, Safety, And Maintenance of Transportation
Facilities", Transportation Research Circular No 370,
May 1991.

89. "Operational Effects of Geometries and Geometric

Design", Transportation Research Circular No 349,
TAB/NRC, April 1989.

90. Owsley C., K. Ball, M. Sloane, D. Roenker, J. Bruni,
"Visual Perceptual/Cognitive Correlates of Vehicle
Accidents in Older Drivers", Transportation Research
Board reprint paper No 910368, TAB Washington, D.C.,
1991.

91. Owsley C., K. Ball, M.E. Sloane, D.L. Roenker, J.R.
Bruni, "Visual/Cognitive Correlates of Vehicle
Accidents in Older Drivers", Psychology and Aging,

92. Owsley. C., & Sloane M.E., "Vision and aging", In F.
Boller & J. Grafman (Eds.) Handbook of
Neuropsychology, Volume 4 (pp229-249). the
Netherlands: Elsevier, 1990.

93.

94.

95.

96.

97.

98.

99.

101.

102.

103.

104.

105.

246

Papacostas C.S., N.H. Kasamoto, "The Intergreen
Interval Controversy: Toward a Common Framework",

Parker, M. and C.V. Zegeer, "Traffic Conflict
Techniques for Safety and Operations", Engineer’s
Guide, FHWA—IP-88-026, FHWA, McLean, Virginia, 1988.

Partyka S. C. Traffic Victim Age and Gender
Distribution. NHTSA, U.S. Department of
Transportation, July 1984.

Partyka 8. Comparison by Age of Drivers in Two-Car
Fatal Crashes. NHTSA, U.S. Department of
Transportation, 1983.

Perchonok K., P.M. Hurst, "Effect of Lane-Closure
Signals Upon Driver Decision Making and Traffic

Flow", Journal of Applied Pshychology Vol 52, No. 5,
410-413, 1968.

Personal Travel in the U.S. In 1983—1984 Nationwide
Personal Transportation Study, Vol. 1, FHWA, U.S.
Department of Transportation, 1986.

Planek T.W. and R.C. Fowler. Traffic Accident
Problems and Exposure Characteristics of the Aging
Driver. Journal of Gerontology, Vol.26, No.2, 1971,
pp. 224-230.

"A Policy on Geometric Design for Highways and
Streets". American Association of State Highway and
Transportation Officials, Washington, D.C., 1984.

"A Policy on Geometric Design of Highways and
Streets". American Association of State Highway and
Transportation Officials, Washington, D.C. 1990.

Ponds R.W.H.M., W. H. Brouwer, and P.C. van
Wolffelaar. Age Differences in Divided Attention in
a Simulated Driving Task. Journal of Gerontology:
Psychological Sciences, Vol.43, No. 6, 1988, pp. 151-
156.

Principles for Urban Planning with Respect to Road
Safety. SCANT, Sweden, 1968.

Prospective Trends in the Size and Structure of the
Elderly Population, Impact of Mortality Trends, and
Some Implications. Current Population Reports,
Special Studies, Series P-23, No.78. Bureau of
Census, U.S. Department of Commerce, 1978.

Rabbit, P.M.A., "An age/decrement in the ability to
ignore irrelevant information", Journal of
Gerontology, vol 20, 1965, pp. 233-238.

Race L., "Traffic Signals", Transportation and
Traffic Engineering Handbook, 2nd Edition, Institute
of Transportation Engineers, Washington, D. C. 1982,
pp 737-782.

 

107.

108.

109.

110.

111.

112.

113.

114.

115.

116.

117.

118.

119.

247

Rainey T.A. and N.H. Pulling, "Performance
Differences on Driving and Laboratory Tasks Between
Drivers of Different Ages", Transportation Research
Record 1281, January, 1990.

Recommended Guidelines for Subdivision Streets — A
Recommended Practice, Institute of Transportation
Engineers, Washington, D.C., 1984.

Retching, 8., Co, J., Fox, M., and Irwin, L.,
"Performance-based measurements among elderly drivers
and nondrivers", Journal of the American Geriatrics
Society, 36, 813-819, 1988.

Rys M., E. Russell, D. Harwood and A. Rys,
"Development of a Headlight Measurement System",
Transportation Research Board reprint paper No
920778, TAB Washington, D.C., 1992.

Rys M., E. Russell, D. Harwood and A. Rys, "Field
Studies with a Headlight Measurement System"
Transportation Research Board reprint paper No
920777, TAB Washington, D.C., 1992.

Salthouse T. A. Adult Cognition: An Experimental
Psychology of Human Aging. Springer-Verlag, New York,
1982.

Saremi A.R., "Identification of the Color of Highway
Signs Under a Combination of Vehicle Headlamp and
High Intensity Discharge Light Sources", ITE
Compendium of Technical Papers, 1991.

Scialfa, C.T., Kline, D.W., & Lyman, B.J., "Age
differences in target identification as a function of
retinal location and noise level: an examination of
the useful field of view", Psychology and Aging, 2,
14-19, 1987.

Sekuler, R., Ball, K., "Visual Localization: Age and
Practice", Journal of the Optical Society of America
A, 3, 864-867.

Senior Driver Facts. California Department of Motor
Vehicles, Sacramento, January 1982.

Shapiro, P., Upchurch, J., Loewen, J., and
Siaurusaitis, V., "Identification of needed traffic
control device research", Transportation Research
Record 1114, 11-20, 1987.

Sharpe, J. A. & Sylverster, T. 0., "Effects of aging
on horizontal smooth pursuit", Investigative
Opthalmology and Visual Science, 17, 1978, pp. 465-
468.

Shinar 0., "Driver Visual Limitations. Diagnosis and
Treatment", NHTSA, U.S. Department of
Transportation, 1977.

120.

121.

122.

123.

124.

125.

126.

127.

128.

129.

130.

131.

132.

248

Shinar, D., "Driver Visual Limitations: Diagnosis and
Treatment", (Contract DOT-HS-5-1275) Washington
D.C.: Department of Transportation, 1977.

Shotton M. A. Belt Up-If You Can. Applied Ergonomics,
June 1985, pp. 127-133.

"Signal Displays For Left Turn Control", Final
Report, FHWA, U.S. Department of Transportation,
1988.

Sivak, M., P.L. Olson and L.A. Pastalan, "Effect of
Driver and Age on Nighttime Legibility of Highway
Signs," Human Factors, Vol 23, No.1, 1981, pp. 59-64.

Solomon D. The Older Drive an Highway Design. In
Needs and Problems of Older Drivers: Survey Results
and Recommendations (J. L. Malfette, ed.), AAA
Foundation for Traffic Safety, Falls Church, Va,
February 1985.

Stamatiadis Nikiforos, "Relationship Between
Accidents of Elderly Drivers and Intersection
Traffic Control Devices", Ph.D. Dissertation,
Michigan State University, E. Lansing, Michigan,
1990.

Staplin L. and R.W. Lyles, "Specific Traffic
Maneuver problems for older drivers", August 1990.

Staplin L. K., M. Breton, S. Haimo, E. Sarber, and A.
M. Byrnes, "Age-Related Diminished Capabilities and
Driver Performance", Working Paper, FHWA, U.S.
Department of Transportation, 1987.

Staplin L., "Age-related Differences in Sensory,
Cognitive, and Psychomotor Test Performance with
Implications for Traffic Control Device Design",
Transportation Research Board Annual Meeting,
Washington D.C., Jan. 1988.

Staplin L., M. Drapcho, "Assessing Age Differences in
Human Information Processing Capabilities Underlying
Traffic Control Device Usage", L. Staplin, M.
Drapcho, Transportation Research Board Annual
Meeting, Jan., 1988.

Staplin L., M. Drapcho, K. Lococo, J. Sim, "Age
Differences in a Visual Information Processing
Capability Underlying Traffic Control Device Usage",
Transportation Research Board Annual Meeting, Jan.,
1988.

Staplin L.K. "Age-Related Diminished Capabilities and
Driver Performance", Working paper submitted by
Ketron, Inc. FHWA Contract DTFH 61-86-C-00044.
Federal Highway Administration. 1990.

Staplin L.K. Age-Related Diminished Capabilities and
Driver Performance. Working Paper. FHWA, U.S.
Department of Transportation, January 1985.

249

133. States J.D. Musculoskeletal System Impairment
Related to Safety and Comfort of Drivers 55+. In
Drivers 55+, J.L. Malfetti (ed). AAA Foundation for
Traffic Safety, Falls Church, Virginia, 1985.

134. Stein, 8.K., ”Risk factors of sober and drunk drivers
by time of day", Alcohol, Drugs, and Drivin , 5(3),
215-227, 1985.

135. Szlyck, J.P., Fishman, G.A., Mater, 8., & Alexander,
K.R., (1990) Visual field screening for driving in
retinitis pigmentosa patients. Investigative
Opthalmology gpg Visual Science (Supplement), 31,
416.

136. Tan C.H., "Increasing Highway Sign Legibility for
Older Drivers", Transportation Research Board
reprint paper No 920711, TAB Washington, D.C., 1992.

137. "Teen and Senior Drivers", State of California
Department of Motor Vehicles, Oct. 1990.

138. "Traffic Control Design Elements for Accommodating
Drivers With Diminished Capacity", Technical Report,
Volume II, FHWA, Washington D.C., May 1990.

139. "Transportation in an aging society",TAB Special
Report 218.

140. U.S. Congress. Senate. Special Commission on Aging.
Aging America: Trends and Projections. 1985.

141. U.S. Department of Transportation. (1985). Survey
data tabulations, 1983-1984 Nationwide Personal
Transportation Study. (DOT-P36-85-1). Washington
D.C.: Author.

142. Untermann R.K. Accommodating the Pedestrian -
Adapting Towns and Neighbourhoods for Walking and
Bicycling. Van Nostrand Reinhold Company, New York,
1984.

143. Upchurch J., "A Comparison of Left Turn Accident
Rates for Different Types of Left Turn Phasing",

144. Van der Zwagg D.C. Induced Exposure as a Tool to
Determine Passenger Car and Truck Involvement in
Accidents. HIT LAB Reports, University of Michigan,
1971.

145. Viano, D.C., Culver, C.C., Evans, L., Frick, M.C. and
Scott, R. (1990). "Involvement of older drivers in
multivehicle side-impact crashes. Accident Analysis
gpg Prevention, 22, 177-199.

146. Walker J., C.A. Sedney, T.M. Mast, "Older Drivers and
Useful Field of View in a Part-Task Simulator",
Transportation Research Board reprint paper No
920702, TAB Washington, D.C., 1992.

250

147. Waller J.A. Medical Limitations and the Elderly
Driver. In Proceedings of the National Conference on
the Aging Driver, American Association of Motor
Vehicle Administrators, Washington, D.C., 1974,
pp.71-84.

148. Waller J.A. Medical Conditions-What Role in Crashes?
New England Journal of Medicine, Vol. 233, 1974, pp.
429-430.

149. Waller P.E., E. C. House, and J.R. Stewart. An
analysis of Accidents by Age. University of North
Carolina Highway Safety Research Center, Chapel Hill,
1977.

150. Waller P.F., E.G. House, and J.R. Stewart. An
Analysis of Accidents by Age. Presented at 56th
Annual Meeting of the Transportation Research Board,
Washington, D.C., 1977.

151. Welford A.T. changes of Performance With Age: An
Overview. In Aging and Human Performance
(N.Charness, ed.) Wiley, Chichester, England, 1985.

152. Willis S.L.. Cognitive Training and Everyday
Competence. In Annual Review of Gerontology and
Geriatrics (k.W. Schaieand C. Eisdorfer, eds.) Vol.
7, 1987, pp 159-188.

153. Wisconsin DOT monograph "Wisconsin's Aging Population
and Their Impact on Transportation", 1986.

154. "Workshop on Roadway Improvements for the Mobility and
Safety of Older Drivers and Pedestrians", summary of
discussions and recommendations sponsored by the
Highway Users Federation, the Automotive Safety
Foundation and FHWA, October 19-20, Arlington
VA,1989.

155. Yanik A. J. What Accident Data Reveal About Elderly
Drivers. Paper 851688. Society of Automotive
Engineers, Dearborn, Michigan, September 1985.

156. Zacks J.L. and R.T. Zacks, "Visual Search Times
Assessed without Reaction Times: Changes with Aging",
Michigan State University,

157. Zwahlen H.T., M. Sunkara, D. Kellmeyer, S. Sundarram,
"Legibility and Conspicuity Performance of Dry and
Wet Beads on Paint and Reflective Sheeting License
Plates under Low beam Illumination at Night",
Transportation Research Board reprint paper No
920865, TAB Washington, D.C., 1992.

158. Zwahlen H.T., U.D. Vel, "Conspicuity in Terms of
Peripheral Visual Detection and Recognition of
Fluorescent Color Targets vs. Non-Fluorescent Color
Targets Against Different Backgrounds During
Daytime", Transportation Research Board reprint
paper No 920880, TAB Washington, D.C., 1992.

 

 

 

 

 

Hui .lr

 

 

 

 

 

‘
.5722 x