You are on page 1of 37

CHAPTER … 5

HIGHWAY ACCIDENTS

"Rest satisfied with doing well, and leaves others to


talk of you as they pleased."

.. Pathagoras (570-495 B.C)

64
5.1 Overview of Road Safety in Developing Countries

Detailed analyses of global accident statistics indicate that fatality rates (per licensed
vehicle) in developing countries are high in comparison with those in developed
countries; countries on the sub-continent and African countries in particular. It is
between 30 to 50 times greater than those in the Western Europe. Furthermore these
studies suggest that whereas in Europe and North America the situation is generally
improving (in reduction of actual numbers of persons killed as well as reduction in
fatality rates), many developing countries have, in recent years, experienced a
worsening situation.

Road accidents in developing countries have been shown to cost on average almost 1
percent of these countries' annual gross national product, utilizing scarce financial
resources that they can ill afford to lose. Studies by U.K. Transport and Road
Research Laboratory, carried out in conjunction with the World Health Organization,
have also shown the road accident ranked surprisingly highly as a cause of death in
developing countries and that the fatality index (i.e. the proportion of all persons
injured who actually die as a result of the road accident can be clearly linked to the
level of medical facilities available. An overall improvement in the number of trained
personnel, hospital beds available or the provision of ambulance services would do
much, therefore, to reduce accident severity.

Although research findings from developed countries can provide some guidance on
the introduction of appropriate road safety measures for developing countries, the
inevitable uncertainties surrounding their transfer emphasize the need for caution. For
example, the problem faced by many developing countries is often markedly different
from that of Western Europe and North America. Thus, a comparison of accidents in
G.B. with six major third world cities indicated that, whereas in G.B. almost two-
thirds of all vehicles involved in accidents were cars, the equivalent value in the third
world cities was often as low as 10 percent. Conversely, the proportions of
commercial and public service vehicles were much greater. In Indian cities, for
example, buses were involved in 25 percent of all injury accidents, the equivalent
figure for Britain being under 4 percent. (This is in part due to the differences in
traffic composition-e.g. the proportion of buses to cars-between the two countries.)

A study of accidents on rural roads in Kenya showed that 16 percent of all road
casualties were occupants of commercial vehicles, with the equivalent value in most
developed countries being under 5 percent. In many developing countries,
commercial vehicles are used to transport people to and from work, and much greater
attention may well need to be paid in developing countries than inn developed
countries to accidents involving commercial vehicles.

These differences in accident patterns, coupled with major differences in road user
behavior, knowledge and attitude, introduce an element of uncertainty in the potential
effectiveness of many types of road accident countermeasure. Results suggest
however, that the introduction of improved education, training and enforcement could
be highly beneficial in developing countries and the potential for improved road
safety by those methods in probably greater than in the developed world.

65
Low cost engineering countermeasures (such as the use of road signs and markings.)
have been shown to be highly cost-effective in G.B. and North America. Little is
known about the effectiveness of such methods in developing countries, but it is
likely that engineering methods that are "self-enforcing" would be much more
effective than those that depend in some way on a reasonable degree of discipline on
the part of the road user.

Scope also exists in developing countries for the introduction of vehicle safety
measures that are both "primary" and "secondary" in nature; the former aim at
preventing an accident, while the latter attempt to protect the road user during the
course of an accident. Much can be done in developing countries to improve the
safety record of public service vehicle, by improving either bus driver training or the
condition and maintenance of the vehicles. Secondary measures, such as the
introduction of seat belts or crash helmets, are non-existent in many developing
countries and efforts should be made to encourage their use.

Almost all countries of the developing world suffer from a lack of financial resources,
and the sums of money available to spend on road safety improvements will be
severely limited. It is essential, therefore, that scare resources are not wasted and that
any measures which are introduced are carefully appraised and an assessment made
of their cost-effectiveness.

This point needs for a good data collection and analysis system. This should be
capable of producing essential information for accident information purposes (at the
national, regional, and local levels) but at the same time it should not be too
sophisticated either for the needs or capabilities of those who operate it or contribute
to it. Recent developments of a system based on a low cost microcomputer have
received much attention in recent years. The use of such a system should enable
countries to invest wisely and effectively in their road safety programs in order to
contain their growing road safety problem.

Finally, the importance of a sound institutional framework should not be


underestimated. It is essential that co-ordination between the various government
departments, monitoring organizations, etc. takes place by means of national road
safety committee. If this is set up at the "highest level (e.g. magisterially or involving
permanent secretaries, director-general, etc.) then a working technical sub-committee
may also need to be established. It is also essential that a small group is set up
(probably within a Ministry of Communications or within National Highways and
Motorway Police) to work permanently on road safety issues and that this important
subject is not dealt with on a part-time basis.

In Pakistan National Highways and Motorways and Motorway Police (NHMP) has
been recently established. Their task is to devise "accident evaluation system (AES)".
Under the system, all black spots will have to be identified on our highways and
Motorways. These black spots will be redesigned with the help of the National
Highway Authority (NHA). The AES has been devised after studying models of
various developed countries, particularly the U.K. Under the system, the data of each
and every accident will be compiled and fully analyzed to look into the cause(s) of
the accident and to suggest the ways and means to reduce the accidents. As per their
information, road traffic crashes killed 1.3 million people every year in the world and

66
injured or disabled as many as 50 million more. Further, the road accidents were the
leading cause of death among young people aged 10-4 years. However, the
effectiveness of the combat for traffic safety will be observed by the experts in the
coming years.

Nevertheless, the limping situation between motor traffic and the present conditions
of roads not improved yet, as well as the low level of traffic morals, the social
tendency of negligence of human lives and inconclusive training for traffic safety, is
the main reason for increasing traffic accidents, which is now taken up as a big social
problem.

As a matter of fact, in most countries public opinion and the official statistics of the
traffic control bodies make a simplified approach to an analysis of the causes of
accidents, most frequently finding their main cause in negligence, mistakes of
motorists or faults in the vehicles. Thus, the World Health Organization considers that
nine of every ten accidents are due to the driver's fault, and that the remaining part
also depends on him to a certain extent. Drivers considered to be responsible for 82%
of all accidents in Pakistan.

5.2 Road Accident Data Analysis Methods

The more developed countries of Europe and North America have long recognized
the importance of a road safety program, but only in the last decade or so as has this
become an area of concern in Pakistan. Accidents in developing countries have very
significant social and economic effects; for example, in a sample of developing
countries, road accidents ranked third behind tuberculosis and dysentery diseases as a
cause of death. The situation for the age group 5-44 years in even worse, with road
accident details ranking higher than any single disease. A major problem when
attempting to deal with the road safety situation in developing countries is the lack of
an adequate statistical data base.

Fatality rates are commonly used to express the scale of the road safety problem and,
as accurate data on total vehicle kilometer are often, unavailable in developing
countries, the most useful indicator is the number of fatalities per 10,000 vehicles
registered. Information from developing countries has shown such accident rates to
be many times higher than the typical values for the countries of North America and
Western Europe as shown in Figure 1. Research has also shown that, whereas rates in
developed countries are decreasing steadily, those in many developing countries are
increasing.

67
The principal uses of a good statistical data base may be defined as:

 The Provision of a general overview of the road accident situation


 The ability to make international and regional comparisons
 The identification of vulnerable groups of road users
 The identification of physical circumstances with high accident levels
 The determination of appropriate national measures
 The monitoring of trends in accident patterns
 The identification of accident black spots
 The identification of appropriate accident countermeasures
 The monitoring of effectiveness of accident countermeasures

5.3 Data Acquisition

The initial reporting of road accidents in virtually all countries swill rest with the
police, although in some countries additional information on the circumstances or
consequences of an accident may be supplied by other bodies, such as highway or
medical authorities. Police will generally be called to all serious accidents and a
proportion of minor ones. The level and time of response will vary with the efficiency
and training of the police officers, resource availability, and the road network
structure. The level of reporting will rarely be consistent and may be more complete
in urban areas with detailed on-the-spot assessment, and will also increase with the
more serious accidents.

An accident should always be defined in purely factual terms, but reporting by the
police inevitably involves a subjective assessment. The factual elements of the

68
accident, such as number, of vehicles and persons involved, maneuvers attempted,
and locations, are required as basic accident statistics. The subjective assessment in
terms of apportionment of blame, necessarily performed by the police, is much more
concerned with prosecution and a clear distinction is required between fact and
interpretation of events. The information required for statistical purposes may be
recorded on a separate form, as in the U.K. or on a combined recording form. The use
of a separate form often results in its completion some time after an accident with
consequential errors. Alternatively, a single form for all accident details may appear
long and over-elaborate.

All data bases will suffer fro under-reporting of accidents as well as errors and
omissions in individual records. A knowledge and understanding of such errors is
essential for proper interpretation of subsequent statistics. The data collection
procedure should be started at the scene of the accident, preferably by the police, with
an easy-to-use accident report form. The form should be designed so that the basic
facts of the accident are recorded whilst subjective questions are minimized.

The requirements of data collection vary from country to country but most will use
the core information shown below:

(i) General accident details

Reference Number
Time
Day
Date
Number of Vehicles
Number of Casualties
Accident Severity
Collision Type

Road Features: Carriageway Type/Junction Type and Control/Speed Limit


Environmental Features: Light Conditions/Weather/Surface Condition

(ii) Vehicle/driver details

Vehicle Type
Vehicle Maneuver
Hit and Run
Driver Breath Test
Driver Age
Driver Sex
Driver Injury Severity

(iii) Passenger/pedestrian details

Injury Severity
Age
Sex
Location

69
(iv) Location details

Region/City Code(s)
Map Code
Grid Reference
Link/Node/Cell Codes
Road Code(s)
Kilometer Post

Accident location is particularly important for cluster analysis, the most common
methods of recording being by:

(a) Grid Co-ordinations: this relies on the availability and accuracy of large scale
maps. The grid co-ordinates will usually be coded back in the office using the
description of the accident scene and the location sketch. For dense urban areas, a
scale of 1: 10,000 is usually sufficient.

(b) Link/Node/Cell: A road network may be easier to define by coding the major
junctions with node codes. Links and cells can be defined using two or more of these
codes. Detailed maps showing the coded network must be drawn up if this method is
to be used.

(c) Kilometer Posts: For an inter-urban highway the accident location is most easily
defined with a kilometer value from the beginning of the highway. Kilometers posts
along highways have proved to be invaluable to both accident analysis and for road
maintenance. Strip maps of the highways showing features such as bridges, villages,
and if they exist, the value of the kilometer posts help to locate accurately the scene
of the accidents.

A location sketch is critical in all of the above methods. It should include the names
of all roads, the location of the accident site to significant landmarks on a north
direction arrow.

5.4 Analysis

For analysis purposes, accidents may be considered as resulting in fatalities, severe


injuries, slight injuries, or "damage only", i.e. no person is injured. Almost all fatal
accidents will be reported, with reduced levels of reporting more minor accidents.
Using only fatal accidents as a statically base will generally not give a sufficiently
large sample size. While the inclusion of serious injury accidents would be likely to
increase the sample size by a factor of between 5 and 10, clear definitions of such
injury accidents are essential. Alternatively, short term assessment of the effect of
remedial measures may be under taken using "conflict" studies in which large sample
sizes are obtained of conflicts which may not result in accidents. However, this
technique requires a substantial team of trained observers.

Manual of the accident data base is possible though rather slow and error-prone.
Simple annual or monthly accident tables are usually the most that are produced. It is

70
therefore preferable to computerize the data base for rapid and accurate retrieval and
analysis of records. The accident report form can be designed so that it is suitable for
the direct input of data into the computer, without the need to transcribe it to a coding
sheet. This is likely to increase both the speed and accuracy of the data entry
procedure. A typical part of such a form is shown in figure 2.

The rapid development of microcomputers has given rise o new opportunities for
developing countries to implement a computer-based accident analysis system. The
transport and Road Research Laboratory has provided further encouragement by
developing a microcomputer based accident analysis package designed specifically
for use in developing countries. It forms, in conjunction with the specially designed
police booklet, a complete system comprising data collection, data entry to a
microcomputer and detailed accident analysis. It is currently under evaluation in a
number of developing countries. The original accident report forms will still play an

71
important part in the detailed analysis as they should contain collision diagrams,
location sketches and witnesses' statements, all of which may be used to supplement
the coded information.

There are two main levels of accident analysis. The first is a national or regional level
from which annual accident statistics may be published, and second is a local level
when individual sites or conditions are being studied.

The cross-tabulation is the most common technique at the national/regional level. Its
uses include identifying individual groups within the population that are most at risk,
identifying physical circumstances with high levels of accidents, monitoring trends in
accident patterns and monitoring the effectiveness of national or regional road safety
measures.

The local level of analysis largely relates to the identification of clusters or "black
spots" and their detailed analysis leading to remedial treatment. For manual analysis,
it is common to produce an accident map in which the exact location of each accident
is plotted using some predetermined coding system of symbols.

For grid co-ordinate system a cell size may be defined (e.g. 100 m 100 m) and the
worst accident sites automatically listed from the computer data base using some
predetermined "trigger" level, or a series of levels, or may be linked to changes in
accident rates. The smaller the cell size the more unique the site becomes so that
eventually it is individual junctions or links that are being listed as, "black spots".
Larger cell sizes are useful for identifying small areas of a region with high accident
densities suitable for "Area Action Plans".

For link/node system and for the inter-urban highways with kilometer posts, it is
possible to rank the worst accident sites or plot histograms of accidents along a route.
In these cases, accident sites might represent a short length of route. In these cases,
accident sites might represent a short length of route (100 m-500 m), a link between
two junctions itself.

Once locations with high accident frequencies have been identified, it is necessary to
consider the accident records in detail to establish common features or patterns. A
"Factor Analysis" may, for example, reveal a predominant vehicle collision type, a
high proportion of night-time accidents or an abnormal number of pedestrian
accidents. The combination of such a factor analysis with a site visit and resource to
the original police report forms will give the experienced traffic engineer an insight to
the causes of the accidents. It is usual for the lowest cost appropriate counter measure
to be implemented and subsequently monitored.

Using low-cost countermeasures it is possible to achieve high economic rates of the


return by subsequently reducing the number and/or severity of accidents at the site.
Assessment will depend on “Before and After” studies in which the changes in
accident rate are compared using standard statistical techniques. The statistical
inferences which may be drawn are very subject to the sample sizes involved. Sample
sizes are in turn dependent on the categories of accident included and the period over
which the analysis is taken. The period of analysis will normally be several years and
comparative statistics from a control area may be necessary to remove the effects of

72
trends which may have occurred on a wider area caused, for example, by national
publicity campaigns or legislation.

5.5 Driving Demands

Driving a vehicle can make certain physical and mental demands on the driver. The
motorists, must for his own and other road-user's safety, make sure that he is as fit for
the road as his vehicle.

Happily no one needed to have the physique of an athlete for handling a family car
efficiently; but a minimum standard of the state of mind and body is essential. Few
motorists would be at ease taking an ailing car on a 100-mile journey. Many,
however, do not care about driving the same distance knowing fully well that they are
not physically fit for it.

Fatigued, uncomfortable, emotionally upset, suffering from a headache or a heavy


cold, dosed with pills and medicines, motorists take to the road seemingly unaware
that even minor ailments can lead to misjudgment, loss of concentration and erratic
performance that may prove fatal.

Motorists who enjoy a general standard of good health, and consider that only serious
ailments like epilepsy, diabetes or a heart disease can cause serious problems while at
the wheels, drive with a false sense of security. Experience shows that the threat to
road safety from minor ailments is far more real involving every motorist at one time
or another, than a killer disease.

No one should drive within three months of an attack of coronary thrombosis.

Warning symptoms, such as double vision or nausea, in people suffering from


migraine must not be ignored by the motorists. They are the signal to stop driving
until the attack has passed.

Basically, all that traffic law requires, as far as fitness is concerned, is that a motorist
fills in the application form for a driving license honestly. The onus is, therefore, on
the motorist himself throughout his driving career to assess his fitness as
conscientiously as he would do far keeping his vehicle road worthy.

If he has any reason to question his ability to drive safely due to certain change in his
health condition, he must consult his doctor. Otherwise, lives of others would be at
stake.

According to law a driver must be able to read in daylight (with glasses, if worn) a
1
vehicle registration number with symbols 3 in. high from a distance of 75 ft (or 3-
2
1
in. high from 67 ft). One-eyed drivers must inform the licensing authority of their
8
defect. A driver must also declare the following disorders and disabilities: epilepsy;
liability to sudden giddiness or fainting; specified mental disorders or defects; loss of
hand or foot; defects in moveable hand or foot; defects in movement, control or

73
muscular power of and arm or leg; deafness and dumbness; any other disease (s) or
disability or liability to make him a dangerous driver.

Although the processes of ageing are in escapable, decreasing the ability to perform
skilled or complicated tasks, age can compensate in the provision of temperamental
balance and experience. The motorist's life may be broken down into six broad age
groups:

 17-21 years: A young driver is inexperienced and most prone to accidents.


Though over-confidence and ignorance he will take risks, thinking "it can't
happen to me." Physical fitness cannot compensate for recklessness. Aggressive
behaviors are common.
 21-25 years: A greater sense of responsibility, more experience and better
judgment lead to safer driving. Sports cars, sometimes rashly driven, have their
greatest appeal. Women drivers are less aggressive, but judgment may be affected
before the menstrual period.
 25-40 years: This is the period during which most people settle down and start a
family life. This leads to more responsible road behavior, with much less risk-
taking. Degenerative diseases, such as ailments of the heart, are rare and there is
little risk of a sudden illness while at the wheel.
 40-55 years: The family man may buy a sports car at this stage in his life. Weight
increases and do the chances of a heart attack. Eyesight may deteriorate.
However, experience makes up for these defects and the middle-aged man is
probably the safest driver on the road.
 55-65 years: Eighty percent of drivers have vision defects by the age of 60. Short
term memory may also be impaired and reaction times slowed down considerably.
Heart attack becomes even more likely. Progressive hearing loss, although not
dangerous in itself, becomes more of a handicap when coupled with failing sight.
 65 years and older: Drivers pass retiring age are faced with degenerative
diseases and slower reactions. Over 70, the motorist should seek medical advice
annually about his fitness to drive. Driving at night on motorways or in large
cities should be avoided.

5.6 Ways of Classifying Accidents

Accidents occur on road sections having different lengths, different horizontal and
vertical elements and differing in the volume and composition of traffic. For this
reason, a direct comparison of different roads with respect to the absolute number of
accidents would give an improper characteristic of safety and traffic conditions.

To obtain comparable data in the analysis of road accidents, a special system of


indices is used, which are called accident rates.

For long sections of roads that are homogenous as regard their geometrical elements,
the accident rate is determined by the formula

z  10 6
Y1  accident per million vehicle-kilometers
365 NL

74
Where
z = number of accidents a year
N = average annual volume of traffic in both directions, taken according to
data of traffic recording
L = length of the section of road, km.

The unit "million vehicle-kilometers" I the above formula, notwithstanding its


seemingly great value, in general corresponds to a comparatively low volume of
traffic. It is not difficult to see that 10 million vehicles-kilometers are performed on a
section 10 kilometers long during about one year of operation with a traffic volume of
3000 vehicles a day. In this case one million vehicles will pass over every bridge on
intersection during this period.

On very short sections of roads that sharply differ from adjacent ones with respect to
the conditions of traffic and the characteristics of the pavement, horizontal and
vertical elements (bridges, road intersections), the accident rate is assessed with the
aid of a different index-the number of accidents per million vehicles using this
section.

The necessity of introducing a second index follows from the fact that the length of
stretches characterized by an increased accident hazard is insignificant, and it is
difficult to establish the boundaries of their influence on traffic conditions, although
there is no doubt that the influence of a short dangerous spot on a road extends
beyond its direct limits. In this case the accident rate is found by the formula

z  10 6
Y2 accidents per million vehicles
365 N

Sometimes, mainly in popular literature, the reverse indices are used-the degree of
safety of travel-showing the number of vehicle-kilometers or vehicles per accident for
a given road section. They have no advantages over the indices described above.

A number of methods have been proposed in foreign countries for a quantitative


characteristic of accidents the introduction of factors or coefficients that take into
account the degree of their danger for road users and the property losses incurred.

In statistics, road accidents are divided into several categories. In many countries
there is a special category of "unreported" or "unregistered" accidents having only
slight consequences, as a result of which the victims were forced to interrupt their
work for less than a week or the material losses were less than some fixed amount
country to country.

The classification of accidents according to their consequences introduces


complications into the comparative assessment of the hazard on separate sections of
roads. In the USA and Germany, the total number of accidents is sometimes assessed
by the severity factor of accidents, calculated by the formula

Y  p1 n1  p 2 n2  p3 n3  p 4 n4

where,

75
n1 , n 2 , n3 and n 4 , are the number of accidents that caused property losses,
light or heavy injuries, or fatalities, respectively.

and
p1 , p 2 , p 3 and p 4 , are the corresponding severity factors for each kind of
accident.

The values of the severity factors, are determined according to an estimate of loses to
the national economy due to road accidents. The difference in values of the factors is
connected with a different assessment of the cost of accidents.

To obtain reliable information on the number of accidents on sections of a road or a


route as a whole, observations must be conducted for a sufficiently long time.
Observations on the London-Birmingham highway showed that the number of
accidents on separate days of the year obeys the Poisson Law. However, Poisson
distribution is applicable when the probability of an event occurring is small, but the
number of possibilities of its occurrence is large, and so it is useful for some of the
problems in traffic. Taking accidents as an example, we know that the probability of
an accident occurring at, say, a cross roads is small. But on a main road the number of
vehicles passing is large, so that some accidents may not be unexpected. The
experimental data deviate from the theoretical distribution since the latter assumes
unchanging traffic conditions. However, the risk of accidents grows on days with bad
weather, while the volume of traffic fluctuates on different days of the week and in
different seasons.

Nevertheless, the traditional method of keeping accident records was to stick pins into
a map. Usually fatal accidents would be marked with a black pin and hence a cluster
of black pins on the map would be referred to as "black spot". Although these pin
maps were religiously kept up-to-date, very little in an engineering sense was done to
try and improve the accident situation in the developing countries until recently when
the accident records were computerized on the federal level.

The computer is also used to identify threshold values of accidents. This involved
printing out all locations with accidents of a certain type over and above some
normally "expected" percentage. Examples of the threshold percentages identified are
as follows:

Driver stopping or slowing down … >60


Driver turning right … >20
Driver doing U-turn … >20
Driver turning left … >30
Driver changing lanes … >20
Light or heavy commercial vehicles … >30
Bus … >20
Fixed object … >20
Same direction-rear end … >60
Same direction-side swip … >35
Turning from wrong lane … >20

76
Opposite direction (as above) … >20
At angle-both traveling straight … >30
At angle-one or both turning … >20
Driver reversing … >20
Wet weather … >20
Children under 16-killed or injured … >10
After dark … >20

Such type of printout yields a wealth of information depending upon the program one
is using. In U.S.A., at the National level there is only one program that collects and
provides information on highway accidents-NHTSA's fatal accident Reporting system
(FARS). The National Safety Council's Accident Facts publication also provides data,
but its scope is fairly limited. In the implementation stage is NHTSA's National
Accident Sampling System (NASS). None of the three systems provide the type of
information that is needed for a comprehensive analysis and evaluation program for
looking at highway environment.

5.7 Fatal Accident Reporting System

The Fatal Accident Reporting System (FARS) is the most comprehensive and detailed
national data file on motor vehicle fatalities. It is the only detailed national data base
system for collecting and providing traffic fatality data.

FARS is a computerized system that contains data on all fatal motor vehicle traffic
accidents. It is operated by the NHTSA's National Centre for Statistics and Analysis
in Washington, D.C. The file was started in 1973 and is updated throughout the year.

Each fatal accident that occurs on a traffic way is reported to FARS.

The source of the input varies form state to state. Typically, State highway department
records, vital statistics, motor vehicle registration files, and police accident reports are
the primary source of FARS data. In fact, NHTSA has contracts with each State and
pays to have State accident analysis review and encode data on special FARS reports
forms. Such data covers the accident, general environment factors; accidents type;
vehicle information; driver-occupant, non occupant; and other details.

The FARS defines a motor vehicle fatality as one that occurs within 30 days of the
accident and is a result of the accident. All other definitions used in FARS conform to
the American National Standards Institute (ANSI) Manual on Classification of Motor
Vehicle Traffic Accidents.

FARS is used by NHTSA to produce special data runs to identify unique areas of
interests.

FARS is limited because it deals only with fatal accidents and cannot be used to
obtain other data to look at the overall accident problem.

The NHTSA's experience with the FARS file since 1975 has resolved many problems
and the FARS system is generally capable of producing useful data.

77
National Safety Council (USA): The National Safety Council (NSC) has been
collecting traffic data since it was formed in 1912. Currently, traffic fatalities are
reported to NSC by the 50 States and the District Columbia. The Safety Council is
confident that they receive most of the fatalities because they believe their annual
reporting correlates very closely with data from the National Centre for Health
Statistics (NCHS).

Annually, the Safety Council produces Accident Facts, a compilation of statistics on a


wide variety of accident types, including traffic data. The International Bridge,
Tunnel, and Turnpike Association; NCHS; the National Highway Traffic Safety
Administration; the Federal Railroad Administration; the Federal Highway
Administration; State and local authorities are frequently quoted in the publication.
Data submitted by States and localities for publication in Accident Facts varies and is
sometimes incomplete so that Safety Council uses projections and estimates when
necessary to establish figures.

The Safety Council data that appears in Accident Facts are fairly limited and not in
the form or data base that can be readily compared to the data collected by the
NHTSA program. Key differences in the Safety Council data collection and NHTSA's
data are as follows:

 FARS deals with detailed factual information obtained for each fatal accident,
whereas the National Safety Council obtains data from States and localities and,
when necessary, uses estimates and projections to establish figures.
 FARS only counts fatal accidents that occur on traffic ways, whereas Safety
Council counts non traffic fatalities.
 FARS uses a 30-day cut off date for individual fatalities (which does not conform
to the ANSI Manual), whereas the Safety Council use a 90-day cut-off date for
fatalities (which does conform to the ANSI Manual).

National Accident Sampling System

In October 1977, the NHTSA introduced a pilot program for the National Accident
Sampling System (NASS) with the establishment of ten accident investigation teams.
They used trained technicians, organized in teams of two to four members, to
investigate a sample of highway accidents in accordance with a predetermined
sampling plan at geographically distributed sites throughout the country. By 1982, 75
teams were established and 17,000 to 25,000 accidents were investigated annually on
a continuing basis. In addition, special studies of accident problems of particular
concern and ancillary studies to gather data, such as exposure information was
performed.

The NHTS stated the following specific objectives for NASS:

1. Estimate and disseminate annual national totals and rates of accidents and
exposure, accident causes and consequences at a level of detail not available in
other systems.
2. Monitor changes and trends in the design of future countermeasures.
3. Evaluate existing countermeasures, motor vehicle standards, and highway safety

78
program standards.
4. Provide data during the field test or demonstration phase of proposed standards
and countermeasures to assist in evaluating their accident and injury reducing
benefit.
5. Provide a current and detailed accident and injury causation data base suitable for
establishing priorities for and assisting in the design of future countermeasures.

The NASS program is of sampling and the data, when encoded into the computer,
could not be related to the site from which they wee derived. When more and better
highway environment data are collected, the system becomes of primary value in
determining overall design criteria. It has no value in identifying site-specific
hazardous locations or for evaluating site-specific improvements. However, its value
for the highway environment improvement could be directed towards data to evaluate
differing characteristics of the facilities.

The NASS program, through specially designed studies is no doubt resource limited.
This particular program is very well in great demand. However, few
recommendations were suggested for the improvement of NASS program. Among the
concerns expressed by the Safety Board was that NHTSA had heavily oriented the
program towards gathering information on the vehicle to the exclusion of the
highway environment. NHTA has agreed to increase the number of data elements
related to the highway environment. It should be realized that any major
improvements will require the support of safety professionals to gain the needed
commitment from government policy makers. This will only be accomplished when
engineers can clearly communicate their specific needs and when purpose of their
activity is understood. Without an understanding of their purpose and needs they can
only expect a dismal future as far as meaningful and top level program support are
concerned.

As concluding remarks, the ways of classification accidents, as discussed above


represent a landmark in change in the safety community's concept of providing a safe
highway transportation system. Safe and efficient highway transportation requires
solutions to complex problems demanding the compatible blending of several diverse
disciplines. As a matter of fact, the accident record is more a symptom, a "side
effect," of the highway transportation problem. Fortunately, most of the
improvements in highway and vehicle design, over the years, have benefited both
transportation efficiency and safety; otherwise macabre statistics would have been
worse. Thus it is noteworthy that auto mobilization of Pakistan should not be
permitted to be attended by a growth in the number of accidents, by the killing and
injuring of people and the loss of material values. It is the duty of all specialists
connected with the solution of problems of traffic safety improvement on roads to do
everything in their power to make roads convenient and safe to travel on. Workers in
the field of designing, constructing and maintaining roads during all their activities
should never forget about ensuring traffic safety. This should be a creative process
based on a deep study of the laws of travel of traffic streams.

5.8 Accident Studies Before and After Changes

To estimate the effect of a change in road conditions at a particular site on the

79
frequency of accidents there, the usual procedure is to obtain details of accidents at
the site in convenient periods before and after and to compare the ratio after to before
with the corresponding ratio for a large control area. The latter may be the whole of
the police district in which the site lies, or some other area from which trends due to
external factors can be reliably assessed. The significance of the difference between
the two ratios can be tested in the usual way by means of Chi-Square Test for
Goodness of Fit with 1 degree of freedom.

For selection and comparison of the effectiveness of possible alternatives of


reconstructing separate sections, the experimental data (RRL, U.K.) given below can
be used:

Improvement Reduction in number of accidents

Improvement of Geometrical Elements of Road

 Complete reconstruction of section


using higher technical standards … 50% and more of total number
of accidents
95% of personal-accident

 Construction of a medium … 30% of same

 Bypassing of small towns and


Villages … 25% of same

 Widening of roadway … 10-15% and more of total


Number of accidents

Reconstruction of Horizontal Curves

 Construction of divisional island … 95% of personal-injury accidents

 Reconstruction with increase in radius… 70-80% of same

 Improvement in sight distance … 65% of same

 Construction of super-elevations
of sharp curves … 60% of same

Improvement of Roadbed and Right-of-Way

 Construction of Sidewalks for


Pedestrians … 35% of total number of accidents
and 30% of fatal ones

 Installation of guardrails … 15-20% of personal-injury


accidents

80
 Removal of lighting standards
from shoulders … 10-12% of total number of
accidents

These data mainly relate to the experience gained in GB and in the U.S.A. They give
only the most general orientation in the effectiveness of measures for improving
traffic safety. In many concrete cases, naturally, the actual may appreciably differ
from the average ones given above.

When simultaneously planning several improvements, one should not expect that
their combined effect will be equal to the sum of the accident reduction figures given
above.

For an objective estimate of the effectiveness of improvement of dangerous sections,


the number of accidents before and after reconstruction is usually compared for the
same number of years.

The reliability of the data obtained is tested by methods of the theory of probability,
on the basis of the assumption that the distribution of accidents on identical sections
obeys the Poisson distribution. It is also assumed that during the comparison, period,
the volume of traffic on the road did not appreciably change.

The task consists in establishing that the reduction in the number of accidents during
a certain period of time after reconstruction meets the same laws of distribution of
random occurrences as before reconstruction.

The task consists in establishing that the reduction in the umber of accidents during a
certain period of time after reconstruction meets the same laws of distribution of
random occurrences as before reconstruction.

For this matter use is made of Chi Square  x 2  test, by means of which it is established
whether the experimental data meet the requirements of the allowable deviation from
the theoretical Poisson distribution curve.

In the simplest case with one group of data before and after reconstruction (one
degree of freedom of Chi Square law), the criterion for the problem being considered
is determined from the expression

( n 1 t 2  n 2 t1 ) 2
x 
2
 x 2 norm
t1t 2 (n1  n 2 )

where,
t1 and t 2 = periods of time before and after reconstruction of a stretch of a
road for which statistical data on accidents are available.

n1 and n 2 = corresponding number of accidents

x 2 norm = minimum values of Chi Square at which the probability of


deviation of the laws of accident occurrence after reconstruction

81
P from the laws existing before reconstruction does not exceed
the permissible values (usually 5%).

The following relationship exists between P and x 2 norm with one degree of
freedom:

P, % … 10 8 5 3 2 1 0.1
2
x norm … 1.71 2 2.7 3.6 4.25 5.41 9.6

Example,
Before reconstruction of an at-grade intersection, there were 17 accidents during 5
years. After reconstruction, there were 03 accidents during two years.

Using the Chi Square test, we have (with P = 5%)

(17  2  3  5) 2
x2   1.8  2.7
5  2(17  3)

Thus, the statistical data available are not yet sufficient for considering with a
probability of 95% that the relation reduction in the number of accidents is due to
reconstruction of the intersection. Assume that one more accident occurred during the
next year. Considering the entire period of operation of the road after reconstruction
of the intersection, we get,

(17  3  4  5) 2
x2   3.05  2.7
5  3(17  4)

Additional analysis confirms the fact that the reduction in the number of accidents by
67% is connected to reconstruction of the intersection.

Furthermore, to review statistically the effect of a change in road conditions at a


particular site on the frequency of accidents there, the usual procedure is to obtain
details of accidents at the site in convenient periods before and after and to compare
the ratio after to before with the corresponding ratio for a large control area. The latter
may be the whole of the police district in which the site lies, or some other area from
which trends due to external factors can be reliably assessed. The significance of the
difference between the two ratios can be tested in the usual way x 2 with 01 degree of
freedom.

As a matter of fact, the chi-square Test (x 2 ) for goodness of fit compares the
observations with some known or suspected hypothesis and the appropriate
expression for x 2 has been given in the following equation:
2
 (Co  E ) 2 
x 2
=S   …..(1)
 E 

In this expression, E is the frequency expected from the hypothesis, and o is the
frequency observed. The quantity given by eqn. 1 has approximately the so called x 2

82
distribution. An effect x 2 is analogous to a ratio between two mean squares, in which
the denominator is obtained from the hypothesis, and so has an infinite number of
degrees of freedom. The numerator may have any number of degrees of freedom,
depending on how it has been derived.

There are many forms of this expression, some of which will be given when the test
described in detail. The function is sometimes described as "chi-square".

It is most important to note this form of the expression for x 2 may only be used for
frequencies, and not for any form of measurement, average, or percentage.

There are many forms of the equation for x 2 , a few of which are mentioned here.
One is

x 2 =S (a  pn) 2 / pn  …..(2)

in which n is as usual the number of observations, p is the expected proportion, and a


the observed number. Another form of this is

 a2 
x 2 = S    n …..(3)
 pn 

a frequent use of x 2 in traffic studies is to find the significance of some change in a


factor before and after some alteration in conditions has been made, making a
comparison with some other place in which no alteration had take place, which will
be called the control. Let the number of factors before the alteration had taken place
be a; the number after the alteration be b; and the corresponding number of factors for
the control, respectively before and after the alteration has been made at the test
place, and for the same period, be c and d, we can then set up a table in the form of as
Table 1.

Table 1. The x 2 Test: Before-and-After Testing at a single site

Before After S( )
Test a b (a+b)
Control c d (a+b)
S( ) a+c b+d h= (a+b+c+d)

We can then calculate from the expression, which is derived from eqn.1

n(ad  bc) 2
x2 = …..(4)
(a  b)(c  d )(a  c)(b  d )

This expression involves the calculation of numerators to a high degree of accuracy if


the figures are at all large, because it is the square of the difference between two large
quantities; although once the square has been done, the rest could be done to a lower
degree of accuracy, since x 2 is only needed to two or three places. There is, however,
another form of the expression for x 2 which can be done wholly on the ordinary
calculator. In this we calculate

83
a b
 A ….. (5a)
(a  c) (b  d )

a c
 B ….. (5b)
( a  b ) (c  d )

And then x 2 = nAB ….. (5c)

However, it is calculated, the table has one degree of freedom, because only one of
the factors, a, b, c, or d can be written in arbitrarily; and then the rest must follow
from the totals.

As an example of this, the accidents on a length of road on which a speed limit was
imposed by reason of a system of street lighting may be used. The limit had
previously been refused by the highway authority. The object is to find out whether
the number of accidents for a three year period before, and for a similar period after,
the limit was imposed, were significantly different. For the control, the number of
accidents in the rest of the country was used. The figures are set out in Table 2, using
the notation of Table 1.

Table 2. The x 2 Test: Accidents Before and After


the imposing of speed limit.

Before After S( )
Test a=12 b=20 (a+b)=32
Control c=4339 d=5340 (a+b)=9679
S( ) a+c=4351 b+d=5360 n=9711

a 12
  0.00276
( a  c ) 4351
From eqn. 5a
b 20 0.00373
 
(b  d ) 5360  0.00097

a 12
  0.375
(a  b) 32

c 4339
  0.448 From eqn. 5b
(c  d ) 9679

B  0.0073

Then x 2 = 9711  0.00097  0.073 = 0.69 from eqn. 5c


0.5 > p > 0.1: not significant

Thus in the three years before the imposition of the limit there were 12 accidents, and
there were 20 in the three years after, while the corresponding numbers in the rest of

84
the country were respectively 4339 before and 5340 after.

It is useful to note, and to remember, that if x 2 is smaller than the number of its
degree of freedom it is certainly not significant. It is here less than one, and in
looking it up in the table we see that then the probability for one degree of freedom is
between 0.5 and 0.1 (nearer the former); so it is clearly no significant. We therefore
cannot conclude from the figures that the speed limit made any difference to the
accident figures on this stretch of road, when allowance has been made for the
accidents in the rest of the country.

We can also solve more complex problems by the use of x 2 . Continuing with the
question of speed limit, it might be urged against any conclusion from the last
analysis that while the speed limit may not have reduced the number of accidents, it
would still reduce their severity. The figures from the stretch of road previously
considered are too small to enable this test to be done, but figures are available for the
accidents in one country for two years before and two years after the imposition of
the general 30 miles speed limit in 1935, taken for the stretches of road on which the
limit was imposed. They are set out in below.

Table 3. Accidents for two years Before and two years After
the imposition of the 30 mph limit in 1935.

Type of
Before Limit After Limit Total
Accident
Fatal 11 8 19
Injury 280 296 576
Non-Injury 371 337 708
Totals 662 641 1303

Here we have two questions to ask. The first is, do the figures shown that the number
of accidents were reduced by the limit? The second is, do the figures show that the
limit reduced the severity of the accidents?

This kind of table is called a contingency table, and if it has i rows and columns it is
called i  j contingency table, so that Table 3 is a 3  2 contingency table.

In table 3 there are three degrees of freedom, because the totals are independent,
having been taken from observed figures, and not at the experimenter's choice. The
first step is to test whether the two column totals differ significantly. That is to say,
we test whether the speed limit can be thought of as having reduced the number of
accidents, though the hypothesis we actually test is that the limit has had no effect. If
this were true, we would expect both totals to be the same. The expression of x 2 to
cover this case is

(a1  a 2 ) 2
x2 = …..(4)
n

in which a 1 and a 2 are the numbers in each class, and n= a 1 + a 2 .

85
applying this to totals in Table 3, we get that

x 2 = (662-641) 2 / 1303=21 2 / 1303 = 0.338

this has one degree of freedom and is clearly not significant as the value is much
smaller than its degree of freedom. This probably is actually between 0.5 and 0.7 and
so the figures show no reason whatever to doubt the proportion, which is that the
speed limit has not affected the number of accident.

We may next go on to examine the alternative proposition that the limit has not
affected the severity of the accidents, and for this now take the totals in Table 3 as
fixed. On the hypothesis, we would expect that of the total of accidents before the
imposition of the limit, 662, the proportion of fatal accidents would be 19/1303, and
so the expected number of fatal accidents would be (19  662) / 1303 = 9.65.
Similarly, the expected number, of injury accidents would be (576  662) / 1303 =
292.65. We then set up a new table, Table 3A, in which we show the expected values
in brackets alongside the corresponding observed ones. It is only necessary to work
out two of the expected numbers, when the rest can be done by subtraction, because
the totals must be equal.

Table 3A. The x 2 Test: Analysis of Table 3.

Type of
Before Limit After Limit Total
Accident
Fatal 11 (9.65) 8(9.35) 19
Injury 280(292.65) 296(283.35) 576
Non-Injury 371(359-70) 337(348.30) 708
Totals 662 641 1303

We now work out another value of x 2 , based on eqn. (1). It will be

 (o  E ) 2  (11  9.65) 2 (8  9.35) 2 (280  292.65) 2


S  =  
 E  283.35 359.70 348.30

296  283.35 (371  359.70) (337  348.30) 2


+  
283.35 359.70 348.30

= 2.22

In this there are two degrees of freedom, because the totals are now fixed, and these
added to the one we have already used in testing the totals make up the three for the
table. For two degrees of freedom the probability of getting this value of x 2 is
between 0.1 and 0.5, and a more detailed table shows that it is in fact slightly greater
0.3. Here again, the figure show no reason to doubt the proposition that the speed
limit did not affect the severity of the accidents either.

In this case no control was used. None is, in fact available, but some British statistics
shows that it is doubtful whether the changes in accidents about the time of the
imposition of the general sped limit were large enough to affect the use of the present

86
analysis.

Never the less, the number of possible uses of x 2 is very large and other methods of
use will be found in various textbooks. However, for better understanding of the
above discussion, the reader should have the fundamental knowledge of the statistical
methods.

The two quotations below aptly describe the fundamental principles underlying
mathematics and statistical methods:

"As far as laws of mathematics refer to reality, they are


not certain, as far as they do not refer to reality."

… Albert Einstein

"A statistical analysis, properly conducted, is a delicate


dissection of uncertainties, a surgery of supposition."

… Morney

5.8.1 Poisson Series

The Poisson (French Mathematician) distribution is applicable when the probability of


an even occurring is small, but the number of possibilities of its occurrence is large,
and so it is useful for some of the problems in traffic. Taking accidents as an example,
we know that the probability of an accident occurring at, say, a cross-roads is small.
But on a main road the number of vehicles passing is large, so that some accidents
may not be unexpected.

It is also found that traffic itself tends to be distributed along the road in accordance
with the Poisson distributed, provided it can run freely, and is not held by traffic
lights, or some form of traffic congestion. This can be of much use in the design of
traffic lights, and similar matters.

As an example of the Poisson series, two counts made in Hyderabad (Pakistan) are
given in Table 4. The counts were made for another purpose, to find the average speed
of traffic over a stretch of road, but they were made in a form suitable for the present
analysis. The observers stationed at each end of the stretch with synchronized watches
and noted the exact time of arrival of each vehicle to one-hundredth of a minute, and
also its registration number. It was thus possible to count the number of vehicles
arriving at the check point in any given interval of time. In the Table, this interval was
taken as two minutes, and the left half of the table, column 2 to 7, headed "Out",
shows eastbound traffic; while the right half, columns 8 to 12, headed "In", shows the
westbound traffic. Column 1 shows the number of vehicles passing in the two-minute
interval.

Table 4. The Poisson Series: Traffic Count on Length of Road

87
Column 2 and 8 show the frequency with which the number of vehicles shown in
column 1 passed during the interval. Thus column 2 shows that in the out "Out"
direction, there were nine cases of five vehicles passing during the two-minute
interval, column 8 shows that in the "In" direction there were twelve such cases The
fx and f x 2 columns have their usual meanings, the number of vehicles per interval, in
column1, being x. These figures are used as before to calculate the mean and the mean
square, the estimate of the variance. These should be equal, and will be seen that in
neither case is there very good agreement. This is not altogether unexpected, and the
goodness, of fit can be tested by means of x 2 test.

The individual expectation can be worked out directly, which is actually done in the
"In" table, column 11, but to illustrate the method the probabilities will be calculated
in column 5. dealing first with "Out" table, we have already calculated x , the mean
number of vehicles passing in the two-minute interval, as 3.429. From this we first
determine the value of e  m . This can be found from Table (Appendix-A), but it may
be helpful to readers to give the method of calculation:

log log e = 1 . 63778


log m = 0.53516
Therefore, log log e m = 0.17294

The antilog of this gives log e m , which is 1.4891, and this is subtracted from zero to
give log e  m , because this term is the reciprocal of e m . The process gives log e  m
as 2.5109, whence e  m = 0.0324.

If pj = probability of j cars per interval, the table of probabilities is then worked out as
follows:

p = e m = 0.0324
p1 = e  m m= p m = 0.1111

88
m2
p2 = e m = p1 m / 2 = 0.1905
2!
m2
p3 = e m = p2 m / 3 = 0.2177
3!
m4
p4 = e m = p3 m / 4 = 0.1866
4!
m5
p5 = e m = p4 m / 5 = 0.1280
5!
m6
p6 = e m = p5 m / 6 = 0.0731
6!
m7
p7 = e m = p6 m / 7 = 0.0358
7!
m8
p8 = e m = p7 m / 8 = 0.0154
m!
m9
p9 = e m = p8 m / 9 = 0.0059
9!
m10
p10 = e m = p 9 m / 10 = 0.0020
10!
m11
p11 = e m = p10 m / 11 = 0.0006
11!
S (p) = 0.9991

We see that each of these terms follows from the preceding one by multiplying by m
and dividing by the number of the power of m in the term needed, since

pj p  p ( j 1)
=
j! j  ( j  1)!
6
m m5 m m m
So that e m  e m  p5  0.1280 = 0.0731, and so on.
6! 5! 6 6 6

After this, the next column, column 6, showing value of E is worked out by
multiplying the total number of frequencies, 63, by the probabilities in turn. In the
“In” table this was done directly by taking p as 63 e  m . In last column of each of
the two parts of the table, x 2 is worked out from its equation (see 5.8.1). Several of
the lines have been grouped together to comply with the requirement that the expected
frequency should be 6 or more. In the connection with the significance of x 2 the
table will have two fewer degrees of freedom than the number of groups used for the
calculation of x 2 , after the grouping mentioned in the last sentence has been done, so
that here both tables will have four degrees of freedom. One degree of freedom has
been lost in calculating the mean m, and another is lost because when one less than
the number of groups has been filled in, the last one follows from the general total.
Here we see that the “Out” table has only a small value of x 2 giving a probability of
between 0.8 and 0.7, and so it is clearly not significant. There is thus no reason to
think that the traffic counted in this direction was inconsistent with the Poisson
distribution.

89
On the other hand, the “In” table has a large value of x 2 with a probability of less
than 0.01, which is highly significant. So that in this direction the flow of the traffic
was not consistent with the Poisson distribution. On examination of the f column,
column 8 of the table, it is seen that there was a deficiency in the number of times
three vehicles passed in the interval, an excess in the number of times either none or
one vehicle passed, and also in the higher numbers. A probable explanation of this is
that the stretch observed is a partly built-up and slightly winding level stretch of road.

The last example merely shows how traffic does, or does not, follow the Poisson
distribution. The next one shows how this information could be put to practical use. It
arose because a scheme was being prepared for improving a complicated junction in a
small town. There was not room for a full-sized roundabout without very extensive,
and expensive, demolition of buildings, but it was possible to provide a small one.
From a preliminary study of the scheme, it looked very much as if it would work if
entry from the side roads could be controlled by traffic lights. There was however, one
feature of the layout which might have introduced the possibility of the roundabout
jamming frequency. That would obviously have been a fatal objection to the scheme.
The layout is shown diagrammatically in Figure 3, on which the entering roads are
lettered from A to F.

Figure 3. Diagram of propped roundabout with its entering roads

Roads B and C are one-way streets in directions shown by the arrows. There is no
difficulty in providing adequate weaving distances for the roads from B round to E,
but there is difficulty if roads A and F both enter the roundabout, because they
converge very sharply. It is possible to run road F into road A as shown in the figure,
and then to control the junction by traffic lights working in conjunction with those
controlling the other roads. But the distance d, between the junction of F and the
roundabout, cannot be arranged to allow more than about five or six cars entering

90
road A in the period when the lights would be set against them to enable traffic from
road F to enter road A.

To find this out, a count was taken of the traffic which entered roads A and F from
roads C and E, which the scheme envisages as being allowed to enter the roundabout
together. The method was the same as in the last example. The recording of the
registration numbers of the vehicles in this case enabled their improvements to be
traced. The result is shown in Table 5. In this table, column 1 shows the number of
vehicles passing in the interval, which was half a minute. Column 2 shows the
observed numbers passing in the interval, and the other columns have the same
meaning as before. It is better to show the working.

Table 5. The Poisson Series: Traffic Count at a Junction

The table shows that the agreement with the Poisson distribution is close, and that the
probability p 5 of five vehicles entering is times in the 115 intervals, about one hour,
which was the busiest hour of the day. This is a risk which could be taken, but there is
the possibility that the traffic might increase. As counted, the mean was 0.95, and so if
the traffic increased by 50%, to 172 vehicles per hour, the mean would become
approximately 1.5, and if it doubled the mean would become 2.0. The expected
numbers were then worked out using these two values for the mean in column 7 and 8
of the table, the probabilities being done in the panel at the bottom right of the table.
Column 7 shows that the probable number of times five vehicles might wait in the
half-minute interval is about 1.6 times in one hour if the traffic increases by 50%, and
column 8 shows that it would be about four times in one hour if it doubled. As it so
happens, the junction is on a temporary diversion of through traffic, pending the

91
construction of a by-pass for the town, and the expectation is that this would be
provided before the traffic doubles, in which case the traffic passing through the
junction would not increase so much. There was, therefore, a fair prospect that the
scheme would work at first and possibly even if the traffic increased by 50%,
although it was doubtful if it would do so if the traffic doubled. Nevertheless, the
analysis should that the idea would be worth looking into further. If the layout could
be altered that six vehicles could wait, or if some form of filter could be provided into
road F, the system could reasonably be expected to work for a number of years; so it
was decided to investigate the matter in great detail.

5.9 Highway Accidents Statistics of Pakistan


In the end, sketchy statistics of highway accidents (2000-2006) as provided by the
NTRC, is given in Table 6. However, no further accident information is available with
them. The table shows province-wise the number of accidents in three categories:
fatal, non-fatal, others. Here, may we share with you as Russian novelist Alexander
Solzhenitsyn (1918-2002) once said: in our country the lie has become not just a
moral category but a pillar of the state. This has universal application and in recent
times more so for our nation. Let us forget this and move to our task.

92
Table 6. Shows Accident Statistics (2000-'06) on Highways in Pakistan

No. of Accidents No. of Casualties


Province Year Non-
Fatal Others Total Killed Injured Total
Fatal
2000 2845 3099 0 5944 3212 7376 10588
2001 2629 3042 0 5671 3272 7214 10486
2002 2565 2712 0 5277 3124 6387 9511
PUNJAB 2003 2344 2709 0 5053 2806 6300 9106
2004 2460 2745 0 5205 3088 6629 9717
2005 2275 2567 0 4842 2732 5758 8490
2006 2170 2552 0 4722 2669 5825 8494
2000 905 826 93 1824 982 1506 2488
2001 984 894 115 1993 1079 1640 2719
2002 904 693 99 1696 976 1236 2212
SINDH 2003 885 673 113 1671 976 1310 2286
2004 861 618 109 1588 943 1334 2277
2005 835 611 106 1552 920 1217 2137
2006 835 631 80 1546 943 1312 2255
2000 748 1957 0 2705 846 3104 3950
2001 659 1839 0 2498 709 3006 3715
2002 649 1854 0 2503 746 2795 3541
KPK 2003 613 1973 0 2586 832 3235 4067
2004 706 2005 0 2711 891 4012 4903
2005 476 1999 0 2475 867 4063 4930
2006 818 2100 0 2918 898 4261 5159
2000 69 146 3 218 78 244 322
2001 43 134 2 179 44 214 258
2002 55 129 1 185 67 240 307
BALOCHISTAN 2003 64 145 1 210 67 216 283
2004 56 142 0 198 59 247 306
2005 88 108 0 196 97 179 276
2006 64 143 0 207 68 229 297
2007 79 123 0 202 43 269 312
2000 116 158 0 274 116 170 286
2001 116 168 0 284 135 233 368
2002 109 151 0 260 109 220 329
ISLAMABAD 2003 125 172 0 297 130 249 379
2004 126 173 0 299 175 267 442
2005 154 160 0 314 177 232 409
2006 138 161 0 299 176 238 414
2000 4683 6186 96 10965 5234 12400 17634
2001 4431 6077 117 10625 5239 12307 17546
2002 4282 5539 100 9921 5022 10878 15900
PAKISTAN 2003 4031 5672 114 9817 4811 11310 16121
2004 4209 5683 109 10001 5156 12489 17645
2005 3828 5445 106 9379 4793 11449 16242
2006 4025 5587 80 9692 4754 11865 16619

The figures for the total accidents show negligible decreases in the accidents during
the periods ending 2001,'02,'03,'05 and negligible increases during the periods ending
'04 and '06. The highest being is 6.2% during the year '04-'05.

93
As regards the fatalities, the figures show a negligible increase for the year ending '01
and '04; while there are negligible decreases for the years '02, and '03. The highest
percentage fatalities is being between '04-'05 period.

The accidents' patterns as discussed above do deduce any particular observations. In


other words, the traffic authorities in Pakistan seems have no special program to
lower down the number of accidents or the casualties.

As citizens of Pakistan we have heard about "traffic weeks" being organized, now and
then, by the authorities. But it seems such cosmetic actions have no significant
influence on the reduction of the highway accidents or the fatalities, leave alone the
injuries caused or the vehicle damage. At this point of time one has to observe the
newly formed National Highways and Motorway Police. However, nothing could be
said at this stage. Public could only hope that this organization will strive hard for the
public safety on our Highways and Motorways.

5.10 Dimensional Interaction

Summing up the causes of road accidents, the conceivable dimensional interaction of


four factors are, i.e. environment, driver, roadway and vehicle (Figure 4). In the area
of analysis method accuracy, data error rate estimation is one of the expected analytic
problems to be overcome. This might include factors such as regression to the mean,
variable trends, unexplained variance, random data noise, and variable inter-
dependence.

Figure 4

The analysis is expected to provide an ability to predict the change in accidents


caused by all conceivable actions involving the driver, vehicle and roadway.
Eventually, it should be possible to predict the outcome of such diverse alternative
actions as a driver education curriculum change, a shoulder-paving project, a change

94
in vehicle weight to horsepower ratio, or any combinations of these and other actions.
We believe we are like animal trappers who set their traps and leave them knowing it
will only be a mater of time before an unwary animal comes along and takes the bait.
When that happens, the trap springs shut with no opportunity for escape and the
animal is caught. The driver, like an animal in the wilderness, is unaware and ignorant
of any danger until it is too late and he or she has neither the time nor space to avoid
the jaws of the trap. The traps are hidden in many ways in the form of conflicting
laws, designs for the average or the 85th percentile and inconsistent standards. When
anything is designed for the average, at least 50% of the people are excluded; when
laws conflict, the people are confused; and when standards are inconsistent, systems
and components cannot work together. As engineers you know that better than anyone
else.

As a matter of fact, the truth of the matter is that every accident is a product of
insufficient time and/or space. We need both time and space to avoid a collision
course, to recover from the unexpected and/or perform that which is required.
Granted, some diseases or such as speeding or sleepy driving may contribute to the
time and space gap. But, the vast majority of traffic crashes occur at relatively low
speeds with healthy, sober drivers at the wheels.

5.11 Public Transport, Goods Vehicles Blamed For 46 pc


Road Deaths

To highlight those collisions which caused road traffic moralities but at the same time
are easily preventable are discussed here. The facts and figures were obtained from
the relevant authorities for the city of Karachi. As regards Karachi, it is a metropolitan
port city of Pakistan, with a metro population of 18,000,000 (city population
15,500,000). The city is spread over an area of 1,362 sq. miles (3,527 sq. km). the
above figures were compiled for the year 2010-'11
(http//www.citymayors.com/statistics/largest_cities_el.html). The characteristics
pattern of its population is its ethnic diversity. People speak different language and
practice different religions. It is connected with other large cities of the nation by
road, railway and air network.

Goods vehicles and public transport were responsible for 46 per cent road traffic
mortalities last year (2010), which witnessed a five percent reduction in fatal injuries
on the road, according to data compiled by Road Research and prevention Centre
(RTIR & PC).

The centre—a public-private partnership of the Jinnah Post Graduate Medical Centre,
Agha Khan University and Hospital and the NED University of the Engineering &
Technology with Indus Motor Company—collects data of the JPMC, AKH, Abbasi
Shaheed Hospital, Civil Hospital Karachi and Liaquat National Hospital—the five
major trauma-receiving hospitals in the city.

According to the centre's data, the umber of fatal injuries caused by goods vehicles
and public transport increased form 7 pc (2009) to 24 pc (2010) and 18 pc (2009) to
22 pc (2010), respectively.

95
However, motorbike continued to be the vehicle most involved in accidents i.e. 57 pc.

The data specified that riders and pillion riders constituted the highest number of
causalities i.e . 55 pc of the total casualty data followed by pedestrian covering 25 pc.

Forty percent vehicles had rear end collision.

The highest number of casualties was reported from the National Highway.

Mauripur Road and the main Korangi Road were found to be the most dangerous
roads for pedestrians while Nawab Sadiq Ali Khan Road and Shahrah-i-Shershah Suri
for riders/pillion.

The highest per kilometer casualties for the vulnerable groups were reported on
Nawab Sadiq Ali Khan Road, which involved riders and pedestrians. The highest
casualties took place is Saddar Town (12 pc) followed by Jamshed Town (10 pc),
Liaquat Town (9 pc) and Korangi Town (7 pc).

The highest casualties of rider/pillion, pedestrians and drivers were reported from
Shahrah-e-Faisal whereas the highest passenger casualties were reported from the
National Highway.

The highest fatalities were reported on the National Highway (76 were fatal out of
total 978). For minor and serious cases, the highest frequency was reported on
Shahrah-e-Faisal (48 were fatal out of 1,762) and Korangi Road (52 were fatal out of
1,321).

On M.A. Jinnah Road 1,404 casualties took place as results of accidents, although
fatalities were comparatively low i.e. 29. Similar is the case with Nawab Sadiq Ali
Khan Road where 12 fatal accidents took place out of 828 casualties. The
superhighway had comparatively low number of accidents resulting in injuries to 484
persons, though 48 people die.

A comparative analysis of the facts collected over the past four years showed that this
as the first time since the launch of the centre that a negative trend in fatal injuries in
traffic accidents had been witnessed. However, there was a 12 pc increase in serious
injuries last years as compared to 2009.

Explaining the reasons for this downward trend, it is said by RTIR & PC said that it
was the result of a number of low-cost measures taken by relevant stakeholders. The
measures included putting up sign boards, reflective stickers and banners etc., at
dangerous stops and making arrangements to reduce speeds.

The CDQK had also taken the responsibility to build pedestrian bridges on Mauripur
Road, though the road came under the National Highway authorities, who had been
delaying the required job for the last two years.

The analysis of two previous years' data also showed that there were 71 pc reductions
in fatalities on Shahrah-e-Shershah Suri and 44 per cent reduction in fatal injuries on
Chaudhry Fazal Ellahi Road.

96
Other roads registering a reduction in fatalities were: Manghopir Road (44 pc
reduction), Korangi Industrial Area (35 pc), Rashid Minhas Road (32 pc).

The roads which witnessed an increase in the number of fatal injuries were: Hakim-
Ibn-e-Sina Road (43 pc); main Korangi Road (36 pc); Nishter Road (33 pc); M.T.
Khan Road (31 pc) and Mauripur Road (21 pc).

According to the 2010 data of the centre, 26,315 accidents were reported at the five
hospitals and the total injuries reported were 31,567; major injuries 73 pc (23,061),
serious injuries 23 pc (7,279), fatal injuries 04 pc (1,227).

Interestingly, data compiled by the city police showed fewer casualties (1,099) and
fatalities (491) in accidents last year, though it was collected from all over Karachi.

According to the RTIR & PC, over 55 pc cases were brought by private means while
38 pc by ambulances. The highest casualty cases were recorded in September 2010
(2,982 of 31,567). Among all the locations, the highest casualties were reported at
middle blocks (69 pc) and intersections).

The percentage of persons wearing helmets was considerably lower than those who
were not wearing helmets. Male casualties covered a high percentage i.e. 85 per cent.

Injury severity with regard to age group the highest portion of recorded casualties
belonged to the 21 to 25 demographics.

The highest rider and passenger casualties were of the age group 21-25. for
pedestrians, the highest casualties were of 0.15 demographics.

The highest casualties were reported during day timings covered 41 pc of the total
data, respectively. The percentage of dusk and dark timings remained the same i.e. 23
pc.

The highest casualties during the dawn, daylight and dark timings were reported on
Shahrah-e-Faisal whereas for dusk timings the maximum cases were reported on
Korangi Road.

Injudicious action on the part of driver/rider and error on the part of pedestrian were
found to be major factors leading to an accident.

The report also highlighted many preventable factors that caused accidents like
foot/scarf/ clothing entangled (718 injuries), overloaded goods or passenger (1,334),
under-construction roads (208), dilapidated road (306) and insufficient illumination
(720), deposit on the road (610) and open manholes (102), etc.

Top five towns with fatal road injuries were Korangi (107), Gadap (98), Saddar (79),
Bin Qasim (75), Gulshan (70) and Kiamari (69).

The National Highway (76), Korangi Road (52), Shahrah-e-Faisal (48), Superhighway
(48) and Mauripur Road (40) were found to be the "Fatal Roads").

97
The report identified dangerous curves and diverging crashes on Shaheed-e-Millat
Expressway, Preedy Street and called upon the CDGK to take safety measures there.
However, CDGK had taken some safety measures on Korangi Naddi road, Mauripur
Road, National Highway, KPT Jinnah Bridge and Karachi Northern Bypass Road.

Pedestrian
In the year 2010, 52 pedestrians were killed in road accidents in Karachi alone.
Insufficient pedestrian bridges and vanishing footpaths, especially over and long
signal-free corridors, seemed as a major cause of the killings of 502 pedestrians in
road accidents in the city last year (2010).

According to the figures compiled by the Road Traffic Injury Research and
Prevention Centre at the Jinnah Post and Medical Centre, 1227 deaths in road
accidents in the city were reported at different hospitals last year and more than 40 pc
of the victims were pedestrians.

Describing the last pace at which the footpaths along thoroughfares are being
vanished or illegally occupied by hoteliers, cabin holders and venders as alarming,
residents and that if officials of the city district government and town municipal
administrations did not take urgent action against the unlawful use of footpaths, the
day was not far when everywhere jaywalkers would be seen running roads, dodging
the fast-moving traffic.

A random survey of various thoroughfares and localities showed encroachment on


footpaths had spread across the metro police, giving one an impression as if the
footpaths were available for businesses and could be grabbed by anyone who was
ready to pay some rent to the city government or the respective town municipal
administration.

While several footpaths along various busy roads and main streets have been occupied
nearby eateries, fast food restaurants, barbecues and ice-cream parlors with chairs and
tables placed on them for serving customers, the recent trend of setting up cabins
portable hoarding signs with bills has left pedestrians with no space at all to walk on.
In this situation, people are compelled to walk on roads risking their lives.

Instead of clearing the footpaths of the encroachment, town officials in some parts of
the city in a bid t widen roads have adopted a novel idea of reducing the width of
footpaths from previously six feet to two feet now. Such narrow footpaths are
common in Saddar, Liaquatabad, Gulberg and Gulshan-e-Iqbal.

The other major cause of road accidents involving pedestrians was insufficient
number of pedestrian bridges, especially over major arteries. Unfortunately, theses at
signal free corridors are scarcely used y the pedestrians. Most of the structures are
awfully risky to use because of their dilapidated condition. Nevertheless, so called
pedestrian crossings are better used by homeless and venders.

Nevertheless, the studies carried out in the west show that overhead pedestrian
crossings are not liked by people to use, especially old, ladies and children.

98
As seen from Figure 5, the most common "preventable causes of road accidents" in
Karachi are grouped under seven broad categories (2010):-

1. Foot / Scarf / Clothing entangled .. 718


2. Overloaded goods or passengers .. 1334
3. Under-construction roads .. 208
4. Dilapidated roads .. 306
5. Insufficient illumination .. 720
6. Deposit on roads .. 610
7. Open manholes .. 102

Thus the fatalities are the highest by vehicle overloading (goods or passengers) and
lowest by manholes left open by the local authorities.

Unclaimed Vehicles
The transport authorities informed that some 102 motorcycles that are parked (Sep,
2011) on the premises of Karachi's Anti-Car-Lifting Cell are on the verge of being
deposited in the Police's central vehicle pool, the “Nazarat”. Theses vehicles are
unclaimed amongst some 800 motorcycles recovered or confiscated by the police
between Jan 1 and April 6. Last year, the Karachi police sent well over 300 unclaimed
motorcycles to the “Nazarat”. The police quote many reasons why vehicles remain
unclaimed: In some cases, owners are not in a position to afford the lawyer's fees,
while in others the court direct the submission of sureties of up to Rs. 20,000/= to
regain possession of motorbikes, and the owners chose not to pursue the case.
Sometimes, the police also find that the given residential addresses and contact
numbers are incorrect.

This issue of unclaimed vehicles—both cars and motorcycles—is present in all


Pakistan's towns and cities. There is little effort by the police to correct this, and there
have been found to be in the use of the law-enforcement. However, part of the
problem is the lack of a computerized database which can help the police easily linked
with the complainant/owner. A vehicle snatched or stolen from one part of the
country, if recovered in another town or city, is unlikely to be matched up with its

99
owner. Given the number of vehicles snatched or stolen in Pakistan every year, there
is a need to develop an integrated databases to trace the ownership of recovered
vehicles. Then through the police or other means, provincial government must set up
a system whereby the particular of unclaimed vehicles, including registration and
chassis numbers etc, are regularly and prominently advertised so that people who
have had their vehicle stolen or snatched can find out whether and from where their
vehicle has been recovered.

100

You might also like