|
The Effect of Cellular Phone Use Upon Driver Attention
by James McKnight
A. Scott McKnight
1991
National Public Services Research Institute
8201
Corporate Drive
Suite 220
Landover, MD 20785
Performed under a grant from the AAA Foundation for
Traffic Safety
One of the most popular innovations in automotive travel
in the past decade has nothing to do with the automobile itself, the people
who drive them, or the roads over which they operate. Rather, it is the
ability to carry on telephone conversations while driving.
What CB radios were to the '70s, cellular phones were to
the '80s. From early 1984, when the first complete systems became operational,
the number of cellular phone users has grown to over two million. By the
mid-'90s, when cellular service will be available throughout most population
centers in the United States, the number of subscribers is expected to grow to
between ten and twenty million.
While cellular phones are really elements of communication
rather than transportation, their potential impact upon the latter is sizable.
The prospect of twenty million drivers having the opportunity to place,
receive, or handle a telephone call while driving is not something easily
ignored.
TABLE OF CONTENTS
EXECUTIVE
SUMMARY
Research has shown that use of cellular phones does not
interfere significantly with the ability to control an automobile except among
the elderly, where potentially dangerous lane excursions can occur. However,
the effect of cellular phones as a possible distraction has not been
investigated.
In this study, 151 subjects observed a 25-minute video
driving sequence containing 47 situations to which drivers would be expected
to respond by manipulation of the vehicle's controls. Each situation occurred
equally often under five conditions of distraction: placing a cellular phone
call, carrying on a simple cellular phone conversation, carrying on a complex
cellular phone conversation, tuning a radio, and no distraction (i.e., none of
the preceding). The radio tuning task was included simply to provide a
familiar benchmark. The degree of distraction was measured by comparing
responses under each distraction with those occurring in the absence of any
distraction. Response was measured in terms of both whether the subject
responded and how long it took (with a time penalty for those who did not
respond at all).
All of the distractions led to significant increases in
both the number of situations to which subjects failed to respond and the time
it took to respond to them. Complex phone conversations created the greatest
distraction and simple conversations the least, with tuning the radio falling
in between. Placing a phone call was no more deleterious than a simple
conversation in causing situations to go unnoticed, but delayed responses to
about the same degree as did complex calls. Relative increase in chances of a
highway-traffic situation going unnoticed ranged from approximately 20% for
placing a call in simple conversations to 29% for complex conversations.
The effect of cellular phone use upon response to
highway-traffic situations was the most deleterious for the older age group
(i.e., 50-80). Overall, the increase in likelihood that some highway-traffic
situation will go unnoticed while calling or conversing on a cellular phone
was (for the older group) about twice that of their younger counterparts.
Older subjects were no more distracted by radio tuning than the middle-age
group (26-49 years) and considerably less than the youngest group (17-25
years). As far as time to respond is concerned, age only effected the placing
of cellular phone calls.
While a cellular telephone conversation is no more
distracting than a conversation of the same intensity with a passenger, the
availability of a cellular phone is almost certain to increase significantly
the number of conversations in general and the more distracting, intense,
business conversation in particular. Older drivers, in particular, should be
cautioned against placing calls
INTRODUCTION
One of the most popular innovations in automotive travel
in the past decade has nothing to do with the automobile itself, the people
who drive them, or the roads over which they operate. Rather, it is the
ability to carry on telephone conversations while driving.
What CB radios were to the '70s, cellular phones were to
the '80s. From early 1984, when the first complete systems became operational,
the number of cellular phone users has grown to over two million. By the
mid-'90s, when cellular service will be available throughout most population
centers in the United States, the number of subscribers is expected to grow to
between ten and twenty million.
While cellular phones are really elements of communication
rather than transportation, their potential impact upon the latter is sizable.
The prospect of twenty million drivers having the opportunity to place,
receive, or handle a telephone call while driving is not something easily
ignored.
Cellular Phones and Safety
Thus far, there is no evidence that the use of earphones
poses a hazard to the motoring public. What makes the paucity of evidence less
than reassuring is the absence of rigorous research in the area. There is
the chance of finding out about the involvement of cellular phones in
accidents from the reports that are ordinarily prepared by police or drivers.
An early study by Brown, Tickner and Simmonds (1969) found that use of the
telephone while driving had the effect upon routine driving skills, but did
impair the perception of gaps in traffic. At the same time, driving impaired
performance on tasks carried on over the telephone. A more recent study by
Stein, Parseghian and Allen (1987) studied lane keeping and found significant
degradation when placing phone calls in straight driving or on curves with
older drivers showing the greater performance degradation. The practical
implications of the variation in lane keeping were negligible except for the
older age group (over 55), where the likelihood of exceeding the lane edge
boundary was over 7%, which the authors translate "to a very high probability
of striking an object outside the traveled lane, and thus a high probability
of accident involvement." The risk of lane excursions and crashes decreased
when the cellular phone was moved from console to dash, although the risk of
crash was not eliminated.
The effect of phone use upon the perceptual responses of
drivers is likely to constitute a greater threat to safety than its
interference with vehicle control. First, perceptual processes play a
far greater role in automobile accidents than does vehicle control. "Improper
lookout" and "inattention" are the two leading contributors to automobile
accidents (Treat et. al. 1977). Second, the extent to which cellular
telephone calling interferes with vehicle control can be reduced by dialing
aides (e.g., speed dial) and by placing calls only when conditions permit
relaxation of vehicle control requirements. The effect of phone use upon
perceptual processes, however, is not so readily ameliorated. There is nothing
that can be done to the phone to lessen the disruptive effect that mere
telephone conversation seems to have on perceptual processes in the Brown et.
al. study. Nor can one arrange to schedule telephone conversations around what
are primarily unpredictable perceptual tasks. Third, among the
population whose driving is most effected by telephone use - older drivers, it
is the perceptual processes that undergo the greatest decline. Research has
shown significant age-related decrement in general attention, selective
attention, attention sharing and spatial judgment.
While Brown, Tickner and Simmonds discovered an effect of
telephone use upon perceptual performance, the scope of that performance was
limited to judging gaps in traffic. Certainly the criticality of this task to
safe driving cannot be questioned. However, other perceptual processes are
equally or more critical. The relation of lapses in visual search (lookout and
attention) to accidents has already been mentioned.
The interference of telephone calling with the perceptual
and cognitive processes involved in driving a car is primarily physical and
has to do with the location of the dialing mechanism, the visibility of the
keys, and so on. The distraction that results from the carrying on the
telephone conversation, on the other hand, is largely mental and is greatly
influenced by the nature of the conversation itself particularly, the amount
of attention it demands. Casual social conversation is probably no more
distracting than talking to a passenger. However, an intense business
conversation could well divert a driver's attention to the point that cues of
potential danger may be overlooked. A survey of cellular phone users conducted
in connection with this study showed that, on the average, 72% of
conversations are for business purposes.
Age Related
Effects
The attentional processes that must be shared when
placing, receiving, or carrying on telephone conversations while driving are
known to be vulnerable to age-related effects. The ability to share attention,
as between the phone and the road, has demonstrated a relationship to age in
studies by Craik (1973), Parkison, Lindholm and Urell, (1980), Temple (1989),
and Ranney and Pulling (1990). Deficiencies in the ability to share
attention have also been found in drivers over-involved in accidents
(Mihal and Barrett 1976, Kahneman 1973). A somewhat less obvious but also
relevant variable would be selective attention, the ability to focus
selectively upon one set of stimuli in the presence of others. This ability
has also been shown to decline with age (Clay 1956, Layton 1975, Rabbitt 1980
and Temple 1989). The studies by Kahneman and by Mihal and Barrett just cited
also found declines in selective attention to be associated with
over-representation in accidents.
Age has evidenced relationships with a number of
psychophysical processes that bear tangentially upon use of cellular phones
while driving. Age-related declines have been noted in information
processing (Braune et. al. 1985; Welford 198 1; Rackoff 1974; and Ranney
and Pulling, 1990),problem solving (Case, Hulbert and Beers, 1970; and
Arenberg 1982) and short term memory (Miller 1979; Welford 1981; and Temple
1989).
Purpose of the
Study
The effect of placing and receiving telephone calls upon
the ability to control the motion of the vehicle and the interaction of this
relationship with age seems sufficiently well-established by Stein,
Parseghian, and Allen to obviate the need for further study. 'The same cannot
be said for the effect of telephone use upon the perceptual and cognitive
aspects of driving, often referred to as vehicle "guidance", in contrast to
the control function. While Brown and Tickner showed that perceptual processes
can be degraded by simultaneous telephone conversation, they did not address
the full range of cognitive and perceptual functions involved in driving; nor
did they examine the important age question.
The purpose of the study described in this report was to
assess the effect of telephone use upon the driver's ability to meet the
perceptual and cognitive demands of the highway traffic environment.
Specifically, it attempted to answer the following research questions:
-
- What effect do placing calls and carrying on
conversations have upon perceptually- and cognitively-mediated responses
to highway-traffic situations?
- How do these effects relate to the complexity of the
conversation?
- How do these effects vary across highway traffic
situations?
- How do any of these effects vary with age?
METHODS
The effects of cellular phone usage upon the ability of
drivers to cope with the perceptual and cognitive demands of driving was
studied by confronting samples of drivers with highway and traffic conditions
calling for certain vehicle control responses and comparing the responses
occurring under ordinary driving to the responses when drivers are handling
telephone conversations.
General
Approach
Any attempt to study the effect of cellular phone use upon
driver cognitive and perceptual processes is challenged by the varied and
unexpected nature of the demands that are placed upon these processes. A truly
empirical assessment of the cellular phone's effects requires a measure of the
driver's ability to meet various perceptual-cognitive demands with and without
concurrent use of the phone. Such comparisons can be made without great
difficulty when the demands upon the driver come from fixed characteristics of
the highway environment, such as intersections or signs along the highway.
But, much of what drivers have to respond to involves the actions of other
road users - a driver who may pull out from a side road, or a child who might
enter the street.
In actual driving, the actions of individual road users
are one-time events and therefore do not lend themselves to comparisons. While
they can and have been deliberately staged for research purposes (McPherson,
McKnight, and Wiedman, 1983), the cost of doing so severely limits the number
of events that can be presented to the subjects and the number of subjects who
can be exposed to the events. Therefore, in situ studies are suitable
for registering only those salient influences that can be counted on to
manifest their effects in small sample studies. The effects of phone
conversations upon driving, on the other hand, are likely to be very subjective
Yet, because of the vast opportunity for such distractions to occur, and
because of the enormous opportunity for injury to arise, the effects can
have serious consequences.
Types of
Distraction
The independent variable under study was distraction.
In this discussion, the term "distraction" refers to a diversion of
attention from driving produced by some situation. The situation of primary
concern is, of course, use of a cellular telephone. The car phone itself
involves minimum distraction. The only time a driver is distracted by the
apparatus is during the act of placing a call. Even when the dialing pad is
placed on the dashboard and cut close to the line of sight, attention must be
diverted from the path ahead. There is evidence that when people focus their
attention upon one stimulus, they may fail to perceive another stimulus
separated from the first by but a few degrees of visual angle. To assess the
effect of placing a call upon driver attention, subjects were required (at
various points of the test procedure) to dial a number given them orally by
the experimenter.
The conversations taking place on the telephone are also a
possible distraction. As we pointed out in the Introduction, what
distinguishes cellular phones from in-person conversations is the higher
instance of calls carried on for business rather than social reasons. It seems
likely that calls involving business would be somewhat more attention
demanding than purely social conversations. To allow differences in the
intensity of conversation to evidence any effects upon degree of distraction,
conversation took place at two levels, casual conversation, in which
subjects talked with the experimenter about a variety of largely
inconsequential topics, and intense conversation in which the subjects
engaged in a set of problem-solving exercises. Testing distraction at two
levels of conversation does not assume that the intense cellular phone
conversations are truly more intense than conversations with passengers -only
that level of intensity is a variable that warrants study.
A distraction with which operation of any in-vehicle
equipment is often compared is that of tuning a radio. The comparison is
typically invited by someone defending introduction of a particular piece of
equipment and using radio tuning as a lawyer might use a legal precedent. It
has been used so often as to become something of a benchmark in studying
in-car distraction. For this reason, it was included among the "distractions"
with which telephone conversations were compared.
To gauge the effect of various acts in distracting
attention, we need to be able to compare them with a condition that offers no
distraction, that is, simply driving the car. The people in this situation
might find things to occupy their attention other than driving, they would be
at least free of any planned distraction.
To summarize, the five conditions creating different types
and degrees of distraction were as follows:
-
- No Distraction -The absence of any planned
distraction
-
Placing a Call -Dialing a telephone number on a
key pad located close to the driver's line of sight
- Casual Conversation - Social chit-chat between
subject and experimenter
Intense Conversation-Subjects solving problems
presented orally by the experimenter
Tuning a Radio - Adjusting a car radio to
pre-determined station
Dependent
Variable
The effect of cellular phone use under study was the
degree of distraction from primary driving tasks. Distraction itself is not
directly observable. It is a hypothetical construct that explains why
performance of some task is degraded in the presence of certain conditions.
'The performance degradation becomes the measure of distraction.
The performance of concern in the present study was the
driver's perception of those elements of the highway-traffic environment that
require the driver to do something. Of primary concern are those situations in
which the driver must do something to prevent an accident. A somewhat lesser
concern to society (but important to the driver) are those responses that
enable drivers to get where they are going. In this study, both were valuable
as indicators of perception and therefore any distracting effect of cellular
phone use.
The extent to which cellular phones become a distraction
can be assessed through measures of response to changes in the highway-traffic
environment that require the driver to do something; such as a car ahead
slowing down or a pedestrian about to step into the street. The presence of a
distraction could be inferred from failure to respond when one would otherwise
do so, or from taking longer to respond. Of course, we could simply ask
drivers if they actually saw whatever it is they were supposed to respond to.
However, in this study, such a response would alert subjects to what we were
looking for and quite possibly change the very behavior we were trying to
measure. Therefore, distraction was measured by comparing vehicle
control responses of drivers to simulated highway-traffic safety scenes.
The scenes that were presented to drivers in the study all
involved situations to which they would normally be expected to respond by
some adjustment to the speed and/or direction of the vehicle. One category of
such situations would be normal responses to changes in the route, such as
turning a corner at an intersection in order to follow a predetermined route.
Another would involve responses to traffic controls, such as, traffic lights
or stop signs. Situations presenting possible danger, when perceived, should
lead to some reduction in speed. Where the probability of actual danger is
relatively small, or the distance to it is relatively large, the normal
reaction is simply to take the foot off the accelerator until the danger
becomes imminent. Where the danger is close at hand, and speed reduction more
urgent, the appropriate response is brake application or, in some cases,
steering the vehicle away from danger. If the use of cellular telephones is
having a distracting effect, the distraction should be apparent in a
difference between vehicle control responses when the potentially
distracting influence of the telephone is present versus the response which
occurs in the absence of any distraction.
The measure of distraction was the difference between
responses occurring when no distracting condition was present and those that
occurred under the four distraction conditions making up the independent
variable. Two response measures were employed:
-
- Response Occurrence - Whether or not the
driver responded
Response Time -How long it took the driver to
respond
These two measures made up the dependent variable under
study. The distraction attributable to any one of the potential Distractors
was a function of the difference between that condition and the no-distraction
condition relative to response occurrence and response time.
Study
Parameters
Of the many variables that might influence the distraction
resulting from cellular phone use the one of most concern was age. Of
course age, as the mere passage of time, would not be expected to influence
anything. However, age has demonstrated a relationship to deficiencies in a
number of mental functions likely to affect distraction including information
processing, attention sharing, selective attention, useful field of view, and
memory. To allow the relationship between age and distraction to be examined,
subjects for the study were recruited from a wide range of age levels; with
quotas established such as to assure sufficient numbers of older drivers to
permit any difficulties associated with advanced years to reveal themselves.
Another variable we might expect to influence the effect
of cellular phones upon any aspect of driving would be experience in
their use. Practice in placing calls could lead to dialing without having to
look, while conversing at length could lead to greater facility in attention
sharing. To permit distraction to be analyzed in terms of experience, an
effort was also made to recruit substantial numbers of cellular phone users
for the study.
Study
Sample
To study the relationship between cellular phone use and
the driver's ability to respond to the demands of the highway traffic
environment, we needed a sample that was generally representative of the
driving population at large with respect to those relationships. Samples of
subjects can be quite different from the at-large population with respect to
many variables, including the variables under study, and still be reasonably
representative with respect to relationships among variables. The only
requirements for entry into the subject pool were experience in driving and
the absence of any known problems that would have adversely affected their
response to highway hazards.
A total sample of 150 was believed necessary to provide a
reliable outcome. To assure that the age distribution was not severely biased
in one direction or another, one-third of the sample was to come from each of
the following age groups: Young (25 and under), mid-range (26-49), and older
(50 and older). Since ages were not known until the subjects arrived, the
division could only be approximated. The final sample included 45 young, 57
mid-aged, and 49 older, for a total of 151 subjects. The mean age of the
sample was 39 years, corresponding exactly to that reported for cellular phone
users by Sextro (1989). In order to permit experience to be studied as a
variable, we established a quota of 50 cellular phone users, a quota that was
met.
Subjects for the study were recruited primarily through
posters placed in neighborhood stores and offices. The announcement offered a
payment of $20 for one half-hour's participation in a study involving
operation of an automobile simulator. Since older drivers and cellular phone
users were likely to be underrepresented in the population reached by the
announcement, additional efforts were made to recruit subjects from these
sources. A route to older drivers was offered by the American Association for
Retired Persons "55 Alive" Driver Improvement Program at which we made
in-person solicitations. To attract cellular phone users, we placed flyers
under the windshields of parked vehicles sporting cellular phone antennas.
Test
Procedure
To study the possible distractive effect of cellular
phones, subjects observed a series of videotaped driving scenes to which they
responded by manipulating a set of simulated vehicle controls. The conditions
under which the activity took place were varied systematically across the five
conditions of distraction mentioned earlier (none, placing a call, tuning a
radio, simple conversation, and intense conversation).
Simulation
The only practical means of presenting large numbers of
drivers with the same array of traffic conditions is through simulation.
Stein, Parseghian, and Allen employed simulation to measure the effects of
phone use upon lane keeping. Because lane keeping requires continuous
interaction between what drivers see and what they do, the simulator was of
the interactive variety.
At the present time, interactive simulators are severely
limited in the complexity that their displays provide. The most sophisticated
type of interactive simulator, and about the only type currently in use,
generates images by means of a computer. Since each image must be individually
programmed, the amount of programming rises sharply with the number of images.
The static aspects of the highway environment-road delineations, highway
structures, traffic controls - are obviously easier to handle than are cars,
pedestrians, and other road users, who not only must be programmed to move but
must move in a way that responds to the acts of the driver. Highways filled
with oncoming, overtaking, and intersecting traffic, and sidewalks teaming
with pedestrian s, are simply beyond the computer power or programming
capabilities of present-day art -at least within the realm of what is
affordable. Yet, if use of cellular phones affects cognitive and perceptual
processes, the effects are most likely to manifest themselves under
conditions of high stimulus complexity. Even distracted drivers are not likely
to overlook a pedestrian entering the street when it is the only thing moving
for miles around.
Simulating the complexities of the highway traffic
environment with any fidelity at all requires the use of cameras rather than
computers. In this study, we videotaped a series of scenes through the
windshield of a moving automobile to create the driving tasks to which
subjects of the study would respond. Each 30 seconds of video tape contained
at least one highway or traffic condition requiring a driver control response
to change vehicle speed or direction.
When played back to subjects, each highway-traffic
condition was made to coincide with a task involving some degree of possible
distraction from use of cellular phones, including no distraction, making a
cellular phone call, engaging in a casual phone conversation and engaging in
an intense phone conversation. The effect of the cellular phone could then
be assessed by comparing responses to the highway, traffic conditions which
arose during calling or conversing on the phone with responses to conditions
that arose when no distractions occurred. A radio tuning task was included as
a reference aide, to allow any distraction associated with cellular phones to
be compared with a form of distraction that i s generally familiar. The effect
of the distraction upon the driver's perceptual and cognitive functioning
would be assessed by studying whether and how quickly people respond to
various highway traffic conditions. The greater the distraction from a cellular phone, the less likely the subject would be to respond or the longer it
would take for response to be initiated.
Driving Scenes
The scenes presented to subjects involved some 47
situations to which drivers might ordinarily be expected to respond. The
situations included the following:
-
- Vehicles -Stopping, turning, entering,
crossing, etc. (18)
Road Configuration - Lane drop, lane
control, narrow bridge, etc. (10)
Pedestrians or Animals - (4)
Route Change - (4)
Road Sight Limitations - (3)
Roadside Construction - (3)
Traffic Control Signal -
(3)
Road Surface Conditions - (2)
A Betacam video camera was mounted in the rear seat area
of a Mazda 626 at about the height of a short driver. The camera was
approximately equidistant between the left and right sides of the car. A
rear-view mirror was placed in the camera' s field of view providing a view of
the area to the rear of the car corresponding to the view normally seen in a
driver's rear-view mirror with the exception that this mirror also reflected a
very small portion of the camera itself. The presence of the camera, however,
had the affect on the ability to see and understand what was happening to
the rear. All routes were identified in advance of the day of shooting as well
as the location and nature of certain pre-planned conditions to occur along
that route. In addition to pre-planned situations there were many
naturally-occurring events that were recorded along the routes. Indeed, most
of the situations appearing in the video involved characteristics of the route
or traffic situations that just happened to arise during the videotaping.
The four route changes were brought about by
superimposition of an arrow in the upper-right corner of the video image. The
arrow pointed upward, indicating no direction change, except at the four
points along the route where a turn was to be made; that is, where a turn
occurred in the video image. Originally, we had intended to plant route sign
facsimiles along the side of the road as a means of guiding subjects. However,
many of the subjects were familiar with the stretches of road employed in the
test, and were aware that no such routes existed. Superimposing an arrow upon
the image allowed us to test the subject's ability to perceive route changes
without requiring them to remember or respond to a designated route.
A 3/4" copy of the Betacam master was played back from
3/4"deck into a 50" screen rear-projection television. Footage was viewed to
identify sections containing a variety of hazards which occurred frequently
and at timely intervals. The "best" sections were edited and spliced together
to create a program that ran approximately 25 minutes. This program was then
screened to identify all highway-traffic conditions that could be used to test
reactions. A pilot group of available & "subjects" drove the simulator
along with the video. The situations that drew reactions from one or more of
the subjects were noted. This information was used to determine which
situations were worthwhile for use in the video. A situation was eliminated
where two situations occurred too closely to one another. This did not allow
sufficient time for the termination of one distraction after the first
situation and the beginning of a new distraction before the next situation
began. This culling process resulted in a list of 47 evenly-spaced situations
to which subjects could be reasonably expected to react.
Response Recording
The perceptual responses to the highway-traffic
environment that might be affected by cellular phones are observable only by
those doing the perceiving. Distraction was therefore measured through
observation of a subject's physical response to the various situations
pictured in the video scenes that were presented to them. As a subject "drove
along" with the video scene, a video camera and VCR recorded the subject's:
Accelerator use, by means of a voltage meter connected to
a potentiometer that was, in turn, connected to the accelerator pedal
- Braking, by means of a light that was connected to a
brake light switch
- Steering and turn signal use which were visible to the
camera
The accelerator meter and brake light were mounted behind
the camera and reflected into the camera's view with a mirror mounted in front
of the camera.
Cellular Phone Tasks
The three telephoning tasks, and the "benchmark" radio
tuning task were controlled by the test administrator in the following manner:
-
- Radio Tuning -To initiate the radio tuning
task the administrator would press a button that turned on a radio next to
the subject. This would be the cue to the subject to turn on the simulator
radio and to try to match the test administrator's station. The radio was
left on until the situation section was over. In the event that there was
sufficient time between the end of one situation and the next, the radio
would be left on a while longer to give the subject as much time as
possible to find the correct station. We felt that subjects who were
successful in finding the station would be less likely to become
discouraged and would therefore take the task more seriously.
Call Placing-To initiate a call-placing task the
administrator would press a button that lit a light just under the TV
screen. This was the cue to the subject to place a call to his or her home
phone number. If it was necessary to extend the length of the call-placing
task to make it last through a longer situation section, the subject would
be told that the line was busy, which was a cue to call a second number,
such as his or her work number.
Simple Conversations-These generally involved
discussions on subjects including, but not limited to, the gathering of
demographic information (age, car phone experience, familiarity with the
route, etc.), what the subject did for a living, what t he subject did with
his or her free time (e.g., what they did during the previous weekend or
what they might be doing after work that evening). It was not generally
difficult to keep an active conversation going throughout the situation
section and terminate it quickly at the end of the situation section.
Complex Conversations-These consisted of either
math problems or short-term memory problems. The math problems consisted of
a string of simple computations (e.g., 2 + 3 + 4 + 1 / 2 x 3 + 4 + 6). Each
computation was simple enough that subjects with limited math abilities
could reasonably be expected to perform them. However, since subjects were
required to keep a running total in their heads, they needed to maintain
attention for the duration of the problem to get a correct answer. In the
short-term memory task subjects were read a list of five or six digits and
were then asked whether certain digits were in that list.
Simple and complex conversations were generally preceded
by a call-placing task which acted as a natural lead-in to a conversation.
Since it takes longer to get into and out of conversations, these distractions
were generally programmed to stretch over two situation sections.
There was nothing administrators had to do to initiate a
no-distraction situation except to assure that the previous distraction was
terminated by the time the no-distraction section began.
Test Forms
The individual subjects could not be exposed to every
highway traffic situation under all five distractions without being exposed to
the same situation five times, an undesirable state of affairs. In order to
limit subjects to a single exposure, the distracting conditions were
introduced in five different sequences such that each sequence matched a
particular highway traffic situation with a different distracting condition.
The different sequences or "forms" were rotated among subjects in such a way
that each form was given equally often. This meant that, across the entire
sample, each distraction condition was paired equally often with each highway
traffic situation.
Administrative Procedure
The test was administered from a room, separated from the
simulator by a twenty foot hallway and a closed door. The test administrator's
main function was to control the presence and type of distractors. The
administrator observed the simulator video via one video monitor and the data
collection video via a second monitor. This made it possible to view what the
subject was watching on the simulator which, in turn, facilitated the timing
of distractors to the appropriate hazards. Monitoring the data collection
video made it possible to assure that the camera was capturing the necessary
information, the assorted meters and lights were functioning properly, and
that the subject was "driving" in a realistic manner. The test administrator
also monitored the beginning and end of the hazard sections via speaker from
the audio track of the 3/4" deck and addressed the test subject via microphone
to a commercial stereo tuner/amplifier which was connected to a speaker in the
testing room. An audio signal on the simulator video tape that identified the
beginning and end of each hazard section was sent to the test administrator to
help the administrator control which distraction was in place during each
hazard. This audio program was also sent to an audio mixer that mixed it with
a microphone in the simulator room and sent it to the data video recorder.
This made it possible for the person reducing the data to time reactions from
the same point for each hazard, for every subject.
A noise gate was connected between the microphone and the
amplifier to turn the microphone off automatically when administrator was not
speaking. This was necessary to prevent the subject from hearing the radio
station being tuned prior to tuning task s and hazard section cues being sent
into the administration room.
The test administrator had an instruction sheet for each
subject which served as a guide to which distractions should be given during
each highway-traffic situation.
-
- Data Reduction
All data reduction was performed from the data video tape,
that is, the video of the subject's performance. Data reducers used a
stopwatch to time the reactions to the hazards and noted cases where the
subject did not react at all. Performance in the distracting tasks was also
recorded. Subjects were scored high on distraction tasks for:
-
- Completing calls quickly
- Searching for radio stations continuously, with no
apparent need to stop tuning to deal with the hazard
- Engaging in simple conversations continuously, with
no apparent need to stop talking to deal with the hazard
- Correctly solving math problems and correctly
identifying numbers that were or were not included in the list of numbers
given in the short-term memory task
Data Analysis
The analysis involved comparing different forms of
cellular phone use and radio tuning with respect to the degree of distraction
in responding to various highway traffic situations. A measure of distraction
was obtained by subtracting an individual subject's "scores" (response time
and proportion of situations responded to) under no distraction from
the scores under each distracting condition. The difference between the two
scores became the measure of distraction. These scores were then aggregated
across subjects, separately for response time and proportion not responded to.
The analysis employed was a factorial Analysis of
Variance, in which the independent variable - the four potential distractors -
formed the factor of primary interest. Two other factors included in the
analysis were age and form. The importance of age as a parameter was
discussed earlier. For any analytic purposes, subjects were divided into the
three age groups noted previously: 25 and under, 26-49, 50 and over. The
oldest age group actually ran up to age 80, with a median age of 61 years.
Test "form" was isolated as a factor for purely
statistical reasons; since the test forms are entirely arbitrary, their
relation to any of the variables in question was not an item of interest.
While each test form consisted of the same high way-traffic situations, the
way the situations interacted with the various distracting conditions was such
that forms might differ somewhat with respect to the two dependent variables.
Since it was not possible to balance the test forms across the different age
groups, failure to control for the effects of test form could end up
introducing a relationship between distractor and age that was truly an
artifact. Treating the test form as a factor prevented any differences in
forms from effecting the other comparisons.
RESULTS
- Effects of Distractions
- Effects Of Age
- Effects Of Experience
- Relative Performance Decrements
- Specific Situations and Distractions
- Performance on Distractors
Effects of
Distractions
-
- Figure 3 displays, for each of the four potential
distractors, the level of distraction with respect to response time and
whether or not subjects responded. The two distraction variables displayed
in the figure are not independent of one another; where subjects failed to
respond to a situation, the maximum response time taken by any subject
exposed to that particular situation under that distraction was entered as
the response time. Had this not been done, the non-responders would not
have app eared in the response time data and the results would have been
meaningless.
FIGURE 3 INCREASE IN REACTION TIME AND
NON-RESPONSES BY DISTRACTION TYPE
Click here for Figure 3
All of the potentially distracting conditions yielded some
degree of distraction, that is, they produced reaction times and non-responses
that were different from the no distraction condition. The overall level of
distraction was highly significant for both non-responses (F = 36.07;
DF = 1,136; P<.01) and for response time (F = 286.75; DF = 1,136;
P<.01) and under all four potential distractors (P = <.01). Overall, the
various distractions increased the length of time needed to respond to highway
traffic conditions by from .4 to .9 seconds, and the proportion of situations
missed entirely from .06 to .09.
When it comes to which condition led to the greatest
distraction, the results varied somewhat from one of the two distraction
variables to the other. Looking at the proportion of subjects who were
distracted from responding at all, the complex conversations yielded the
greatest interference, while placing calls and carrying on simple calls
yielded the least interference and tuning the radio fell somewhere in between.
The differences among all distractors were only marginally
significant(F=2.133;DF=3,1 34;P=.10). However, complex conversations were
significantly more distracting than simple conversations (F = 4.12, DF = 1,
134; P =.04).
Turning to the time it took to respond, we see that
placing a telephone call rose from one of the least distracting to one of the
most distracting conditions. The differences across distractions are
statistically significant (F=4.37;DF=3,134 ;P<.l0). Considering that those
who failed to respond are included within the response times, it is clear that
it is the delay in responding among those who actually responded that account
for the difference in outcomes. What the results seem to say is that the act
of placing a cellular phone call may be no more distracting than carrying on a
casual conversation in so far as noticing highway traffic conditions is a
concern. However, it does seem to extend somewhat the delay in responding.
When a non-urgent situation arose while a call was being placed, many
subjects delayed responding until they had completed the call. But they did
respond, indicating that the situation had not gone unnoticed.
Effects Of
Age
FIGURE 4 INCREASE IN REACTION BY AGE AND
DISTRACTION TYPE
Click here for Figure 4
Figure 4 displays the proportion of drivers failing to
respond to highway traffic conditions as subdivided by age. It is evident that
drivers in the over-50 category show strikingly higher proportions of failing
to respond to highway traffic situations. The overall effect across
distraction conditions is not statistically significant (F = 2.22; DF = 2,136;
P. 1 13). However, the deficiencies of older drivers significantly exceed those of the other two age groups in telephone calling (F = 7.96; DF = 1, 14 1; P
<.01), and simple phone calls (F = 5.13; DF = 1, 14 1; P <.05), but not
complex phone calls (F = 2.34; DF = 1, 14 1; P =. 13). Also, in tuning the
radio, age differences were not statistically significant (F =.73; DF =
1,141; P =.39).
Part of the explanation for the failure of the radio
tuning task to show significant age effects is the relatively high degree of
distraction evidenced by the 17-25 year age group. The results suggest that
this age group is somewhat more preoccupied with tuning the radio than with
telephone calls, a hypothesis that most parents having children in this age
group would have little difficulty accepting. But why significant age
differences didn't appear in complex cal s lacks a ready explanation. It may b
e that complex conversations are more or less equally distracting to everyone,
while placing calls and carrying on simple conversations only distracts the
older subjects. Perhaps a more parsimonious explanation is that age amplifies
the effects of all telephone-related distractions and that the differences
among the three types of distractions are largely the result of chance.
Turning from whether drivers respond to how long it takes
them to do so, Figure 5 shows the effects of age to be somewhat attenuated.
Over all distraction conditions, the effects of age are statistically
non-significant (F = 1. 14; DF = 2,136; P = < ;.32). The only two
conditions showing a marked increase in reaction time for the older age group
are placing telephone calls and carrying on simple conversations, of which
only placing calls achieves significance (F = 3.01; DF = 2,136; P =.05). The
effect of phone use upon older drivers seems more to prevent them from
noticing various highway traffic conditions rather than to retard their
response to them.
FIGURE 5 INCREASE IN RESPONSE TIME BY AGE
AND SITUATION TYPE
Click here for Figure 5
Effects Of
Experience
Prior experience with cellular phones appeared to have no
significant effect upon distraction resulting from phone use or tuning the
radio. Across all distractions, differences between experienced and
inexperienced subjects were statistically non significant for response time
(F=1.55; DF=4,114; P=<.19), or for the likelihood of responding at all
(F=0.39; DF=4,114;P=<.81). What slight differences occurred seemed to favor
the inexperienced, although such differences, if they exist, can be attributed
to the fact that the experienced subjects tended to respond more quickly when
there was no distraction and might therefore tend to evidence a slightly
greater difference between the undistracted and distracted conditions. In
looking simply at raw reaction times under the various distractions, the
experienced subjects responded as quickly or more quickly than the
inexperienced subjects. In any case, it is clear that prior experience with
cellular phones has no real impact upon the degree to which one is distracted
by its use.
Relative Performance
Decrements
The decrements in performance that have been discussed
amount to greater response time and the probability of not responding as
compared with the results obtained in the absence of any distracting
condition. Just how bad these decrements are can only be understood in
relation to just how slow or unlikely to respond people are in the absence of
any distraction. For comparison purposes, it is necessary to know that the
mean response time in the absence of any distraction (across all highway
traffic conditions) was 4.45 seconds, across all situations, while the
proportion not responding at all was .343, again across all situations.
Considering the proportion of subjects not responding,
the relative decrements experienced by the older age group in placing
calls was (. 127/.343 = ) 37%, simple telephone conversations (.108/.343 =)
31%, and complex phone conversation (.123/.343 =) 36%. For the other two age
groups, performance decrements were much smaller, the largest being a
(.072/.343 = ) 21 % greater probability of not responding for the 17-25 year
age group when making complex phone calls.
The condition leading to marked increases in response
time was where the oldest age group had an increase of 1.417 seconds in
placing calls. Expressed as a percent of the response lag under no
distractions, this translates to increase in response time of 32%. Decrements
in the remaining cases were considerably smaller, falling largely between .4
seconds (9%) and .8 seconds (18%).
Specific Situations and
Distractions
-
- The effect of using the telephone or tuning the radio
upon response to highway traffic situations was not uniform across all
situations. Interaction between the effects of distractions and various
highway traffic situations was evident as a highly significant difference
across the five "forms" i.e., the ten combinations of distractions and
conditions occurring in the video. Recall that five different forms were
needed to allow each of the five phone conditions to be matched with each
of the highway traffic conditions without exposing the same subject to the
test route more than once. Since the forms do not differ with respect to
either distractors or highway traffic situations but only in the way they
were combined, the significant differences among forms means that certain
combinations of the two variables were particularly problematic.
To see if there was any pattern to these aberrant
combinations of potential distractors with highway traffic conditions, they
were examined individually. Specifically, those combinations leading to
proportion of non-response that were discrepant from what would be expected
from the effects of the distractors or highway traffic conditions alone were
identified through a logit analysis.
The results were not at all revealing. The number and
nature of aberrant combinations followed a chance pattern. As to the number,
only four of 235 combinations fell beyond a .05 confidence interval around
the expected results, whereas one would have expected (235 x .05 =)
almost 12 by chance alone. As to the nature, no logical pattern could be
discerned in the results. It should be noted, that with 150 subjects and
five conditions, each condition was only replicated 30 times for a
particular highway traffic situation.
Performance on
Distractors
-
- Thus far, our concern for the effect of various
potential distractors upon response to highway-traffic situations has been
limited to whether or not simply engaging in the task influenced driving
performance. The distracting effect of cellular phone use or radio tuning
tasks upon the response to highway-traffic conditions might be expected to
vary as a function of the amount of attention devoted to the tasks.
A measure of the amount of attention paid to the distracting tasks would
be performance on those tasks themselves. This aspect of performance was
assessed as follows:
-
Radio Tuning -Whether the tuning process was
continuous or whether it was interrupted by the associated highway-traffic
situation
Placing Calls - Length of time required to
complete placing the call
Simple Conversation -Any interruptions in the
conversation coincident with appearance of a highway traffic situation
Complex Conversation - Incorrect answers to the
problems being solved
Time to complete the radio tuning task could not be
used as a criterion since it was largely determined by how much the dial
had to be manipulated to reach the target station, something that varied
by chance from one trial to another.
If differences in quality of performance on the
distracting tasks influenced responsiveness to highway traffic conditions,
chance differences in quality of performance could obscure relationships
under study unless a it was used as a covariate when analyzing those
relationships. However, when quality of performance on the distracting
task was compared to amount of measured distraction, no significant
relationships materialized. For example, whether or not subjects answered
to problem solving questions correctly during complex conversation was
unrelated to the distraction the problem solving caused itself.
DISCUSSION
-
- The three tasks associated with use of cellular
phones -placing calls, simple conversations, and complex conversations
-all led to significant increases in time to respond to highway traffic
conditions and in the likelihood in failure to respond at all. As might be
expected, complex conversations involving problem solving led to the
greater degree of performance decrement - about on par with tuning a
radio. The act of placing cellular phone calls yielded increases in
response time similar to that of complex conversations, but increases in
non-response that were similar to simple conversations.
The overall results conceal large age differences. The
proportion of drivers age 50 and over failing to respond to highway traffic
conditions while using cellular phones was two to three times greater than
that of younger subjects. Among those responding, the oldest subjects took
significantly longer to respond than their younger counterparts when placing
calls, but evidenced no slower response time than the two other age groups
when conversing on the phone. Tuning the radio, while a highly distracting
task, appeared equally so for all age groups. Prior experience with cellular
phones appeared unrelated to the degree of distraction involved in using
cellular phones.
Magnitude of
Problem
How concerned should we be with the distraction created by
use of cellular phones? Of the two dependent variables, non-response and
response time, the former is certainly the more important. First, whether or
not drivers notice and respond to elements of their highway-traffic
environment is certainly more important than how long it takes them to do so.
We are not dealing with emergencies, where time is of the essence. The
situations to which subjects were expected to respond became evident almost
five seconds before the average subject felt it was necessary to do anything.
The decrements of less than a second that result from use of cellular phones
represent a relatively small increment in total response time. Second, the
response time measure employed in the present study was somewhat artificial,
including what amounted to a penalty for failing to respond.
For the driving population at large, simple casual
conversation seems to have little impact upon the ability of people to notice
and respond to the demands of the highway and its other users. Nor does the
act of placing calls seem to divert attention, although drivers may take some
fraction of a second longer on the average to respond. It is those
conversations that require intense concentration on the part of the driver
that appear to be most distracting. When confronted by those highway-traffic
situations presented in this study, their chances of not responding increase
by almost .10, which is approximately a 30% increase over the non-response
rate when no distractor is present. An increase of this magnitude and the
chances of not noticing something, while small, is nonetheless cause for
concern. Someone might point out that the performance decrement it represents
is no worse than that which occurs when tuning a radio. However, the amount of
time during driving that is devoted to tuning a radio may be considerably less
than the time spent in intense phone conversation by those who use cellular
phones for business purposes.
The greatest deficit in ability to respond to
highway-traffic situations is experienced by the older drivers. The frequency
of non-response was from almost two times to over three times that evidenced
by their younger counterparts. The degree of deficit was rather similar across
the three phone tasks (calling, simple, and complex conversation), increasing
the likelihood of non-response by. 1 1-. 13, representing a 33-38% increase
over non-responses in the absence of any distraction. Among the older drivers,
the distraction resulting from the use of cellular phones was again half as
large as that involved in tuning a radio, which was actually the most
distracting for the youngest age groups.
One legitimate question might be to what extent the
distractions from casual and complex conversations are truly a cellular phone
problem. While placing a call is a phone-specific task, the carrying on of
conversations is not. Under the "hands off type of cellular phone simulated in
the present study, conversations were really no different from those that
might be carried on with another passenger. But, what a cellular phone can do
is bring into the vehicle conversations that are more frequent and more likely
to be intense than those that would occur with passengers. From accident
statistics we know that drivers are unaccompanied about two-thirds of the
time. lit seems very likely that introduction of a cellular phone brings about
a significant increase in the likelihood of intense phone conversations.
Implications
The results of the study that has been described carry two
significant implications for use of cellular phones. First, all users of
cellular phones should be advised not to engage in intense phone conversations
while the vehicle is moving. Businesses whose employees regularly carry on
transactions by means of cellular phones might advise, or even direct that
protracted dealings over the phone be avoided while the vehicle is underway.
The second implication has to do with older drivers. Not
only is the performance deficit of drivers over 50 years of age significantly
greater than that of younger drivers, but it prevails over all three of the
cellular phone tasks studied. If there is any group that should not be using
cellular phones while driving, it is those in the older age group.
Gerontological research in general shows that the severe deterioration in
mental processes tends to become more and more prevalent beyond age 70. One
might therefore expect markedly greater instances of failure to respond to
highway-traffic situations at these advanced years. Unfortunately, the number
of test subjects in this age category was not sufficient to permit this
possibility to be tested.
There is no reason to discourage older drivers from having
cellular phones in their vehicles. Phones provide them with a very valuable
way of summoning help in the event of illness or mechanical breakdown without
advertising their plight over a CB radio. However, the diversion of attention,
coupled with the difficulties in vehicle control found by Stein, Parseghin and
Allen (1987) contraindicates their use while the vehicle is in motion.
Conclusions
From the results of the study that has been described in
this report, the following conclusions may be offered.
-
-
- 1. All forms of cellular phone usage lead to
significant increases in the establishment of non-response to
highway-traffic situations and increase in time to respond.
2. Complex, intense conversation leads to the greatest
increases in likelihood of overlooking significant highway traffic
conditions, and the time to respond to them. The distracting effect is
similar to that of tuning a radio. The effect of placing calls or engaging
in casual conversation is less of a problem, although, calling tends to
retard responses.
3. The distracting effect of cellular phone use among
drivers over age 50 is two- to three-times as great as that of younger
drivers and encompasses all three aspects of cellular phone use - placing
calls and carrying on simple and complex conversations. The effect is to
increase non-response by 33-38%.
4. Prior experience with cellular phones appears to
bear no relation to the distracting effect of cellular phone use.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the assistance
provided by Julian Urquijo, who helped assemble the test equipment and
administer tests, to A. Scott Tippetts, who conducted all the statistical
analyses, to Marcia L. Williams, who prepared and edited this report, and to
Shirley Koelsch and Leslie McKnight, who assisted in data reduction.
REFERENCES
Arenburg, D. (1982). Estimates of age changes on
the Benton visual retention test. Journal of Gerontology, 37, pp.
87-90.
Braune, R., Wicens, C. D., Strayer, D., and Stokes,
A. F. (1985). Age dependent changes in information processing abilities
between twenty and sixty years. In Proceedings of the Human Factors Society
- 29th Annual Meeting, pp. 266-330.
Brown, 1. D., Tickner, A. H., and Simmonds, D. C. (1969).
Interference between concurrent tasks of driving and telephoning. Journal
of applied Psychology, 53 (5), pp. 419-424.
Case, J. W., Hulbert, S., and Beers, J.
(1970). Driving ability as affected by age. (NTIS PB 193-927).
Springfield, VA: University of California, Institute of Transportation and
Traffic Engineering.
Clay, H. M. (1956). Age difficulty in separating spatial
and contiguous data. Joumal of Gerontology, 11, pp. 318-322.
Craik, F. (1973). Signal detection and analysis of age
differences and divided attention. Montreal, (Paper presented at the
Annual Meeting of the American Psychological Association).
Kahneman, D. (1973). Attention and effort.
Englewood Cliffs, New Jersey: Prentice-Hall.
Layton, B. (1975).
Perceptual noise and aging. Psychological Bulletin, 81.
McPherson,
K., McKnight, A. J., and Weidman, J. R. (1983). Supplemental
driver safety program development. Vol. I. Developmental research and
evaluation (Final Report on NHTSA Contract No. DOT-HS-9-02284).
Mihal, W., & Barrett, G. (1976). Individual
differences in perceptual information processing and their relation to
automobile accident involvement. Journal of applied Psychology, 61 (2),
pp. 229-233.
Miller, E. (1979). Memory and aging. In M.M. Gruneberg and
P.E. Morris (Eds.), Applied problems in memory. New York: Academic
Press.
Parkinson, S. R., Lindholm, J. M., and Urell, T.
(1980). Aging dichotic memory and digit span. Joumal of Gerontology, 35.
Rabbitt, P. (1980). A fresh look at changes in reaction
times in old age. In D.G. Stein (Ed.), The psychology of aging. Problems
and perspectives. New York: Elsevier.
Rackoff, N. J. (1974). An investigation of
age-related changes in driver's visual search patterns and driving performance
and the relation of tests of basic functional capacities. Unpublished
doctoral dissertation, Ohio State University.
Ranney, T. A., and Pulling, N. H. (1989). Relation of
individual differences in information-processing ability to driving
performance. In Proceedings of the Human Factors Society - 33rd Annual
Meeting, Human Factors Society, pp. 965-969.
Sextro, D. (1989). Analysis: Growing the typical cellular
user. Cellular User.
Stein, A. C., Parseghian, Z., and Allen, R. W. (1987). A
simulator study of the safety implications of cellular mobile phone use. In
31st Annual Proceedings of the American Association for Automotive
Medicine.
Temple, L. L. (1989). Perceptual and cognitive factors
affecting driving ability of young and older drivers. Las Vegas:
University of Nevada.
Treat, J. R., Tumbas, N. S., McDonald, S. T., Shinar, D.,
Hume, R. D., Mayer, R. E., Stansifer, R. L., and Castellan, N. J. (1977).
Tri-level study of the causes of traffic accidents, Vol. I. Causal factor
tabulations and assessments, Vol. II. Special analyses Final Report on
U.S. Department of Transportation Contract No. DOT-HS-034-3-535-77).
Washington, DC: Government Printing Office (Indiana University, Institute for
Research in Public Safety.
Welford, A. T. (1981). Signal, noise, performance, and
age. Human Factors, 23, pp. 97-109.
Revised 7/26
|
|