Available via license: CC BY 3.0
Content may be subject to copyright.
IOP Conference Series: Earth and Environmental Science
PAPER • OPEN ACCESS
Relationship of Accident Rates and Road Geometric Design
To cite this article: Md Hasibul Islam et al 2019 IOP Conf. Ser.: Earth Environ. Sci. 357 012040
View the article online for updates and enhancements.
This content was downloaded from IP address 154.16.56.150 on 26/11/2019 at 03:27
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd
Sustainable Civil and Construction Engineering Conference
IOP Conf. Series: Earth and Environmental Science 357 (2019) 012040
IOP Publishing
doi:10.1088/1755-1315/357/1/012040
1
Relationship of Accident Rates and Road Geometric Design
Md Hasibul Islam*, Law Teik Hua, H. Hamid and Arash Azarkerdar
Department of Civil Engineering, Faculty of Engineering, Universiti Putra Malaysia,
43400 UPM, Serdang, Malaysia
*hasib9898@gmail.com
Abstract: Road safety issue has become a major concern worldwide due to the serious
consequences of road accidents on countries socio-economics and human lives. The main aim
of this paper is to discuss and review the effects of geometric design elements on road accidents
including statistical models used over the years and to compare the outcomes of the studies
conducted in several countries. A way to improve road safety is improving road geometric design
to mitigate accidents occurrence and severity on roadways. In order to improve the road design,
it is crucial to evaluate and define the relationship between road geometric design elements and
road accidents. Studies have tried to relate road geometric design elements such as lane number,
sight distance, super-elevation, median width and type, lane and shoulder width, curve radius,
gradient, and horizontal and vertical alignments to accident rates. Due to the interrelations
between geometric design elements with each other and with other road accident factors, and
also lack of reliable methodology of relationship estimation between road geometric design and
safety. Ongoing enhancement of statistical methods has resulted in many road design-accidents
analysis models which have been developed using a variety of statistical modelling approaches
such as linear regression models, multiple linear regression models, Poisson regression models,
Poisson-Gamma models, negative binomial models, bivariate and multivariate models,
generalized estimation regression models, random parameters models, etc. However, some
limitations have been encountered using the mentioned statistical models, thereby there are
opportunities for road safety researchers to overcome the limitation challenges.
1. Introduction
Road safety has become a major concern affecting countries’ socio-economics. According to World
Health Organization (WHO, 2013) statistics, there are approximately 1.25 million fatalities and 20-50
million who are seriously injured and living with long-term disability every year around the world due
to road accidents. Road accidents are considered as the ninth prime leading cause of death among youth
and expected to become the fifth by the year 2030. Majority of road accidents occur in developing
countries costing USD $ 518 billion worldwide per annum. There are three major factors that attribute
to road accidents namely human-related factors, vehicle-related factors, and roadway- related factors
(William Haddon, 1970). As stated in Highway Safety Manual published by American Association of
State highway and Transportation (AASHTO) (2008), roadway factors contribute by 3% of road
accidents, whereas 34% is attributed by a combination of roadway-related factors and other factors. The
contribution percentage of each factor is shown in Figure 1. Since that the majority of road accidents
factors are interrelated, it becomes complex to identify the certain cause of road accidents as the
accidents occur due to a combination of several factors. Although human-related factors significantly
contribute to road accidents, direct control and prediction of human factors are difficult. Thereby, the
Sustainable Civil and Construction Engineering Conference
IOP Conf. Series: Earth and Environmental Science 357 (2019) 012040
IOP Publishing
doi:10.1088/1755-1315/357/1/012040
2
human factors can be indirectly controlled and predicted through investigation of roadway factors
particularly roadway geometric design. Speed is a critical parameter in road geometric design and related
to road safety. Roadway geometric design elements which have been investigated to influence road
accidents include horizontal curvature, vertical curvature, grade, number of lanes, lane width, shoulder
width and type, median width and type, number of access points, curve radius,
Figure 1. The contribution of accidents factors in percentage (Source: AASHTO, 2008)
Super-elevation, sight distance, and pavement conditions, however some of these elements may not have
a significant impact on road accident rates (Douglas et al, 2000).
2. Review of Models Used to Quantify and Predict Accidents Rate Related to Road Geometry
Several Statistical models were developed to quantify the relationship between road geometric elements
and road accidents, such as linear regression, Poisson models, Poisson-gamma models, negative
binomial models, generalized estimating equation, random-parameters, and bivariate and multivariate
models. Linear regression models were considered and developed in earlier studies to find the
relationship between road accidents and road geometry elements, however it was reported that using
conventional linear regression models showed unsatisfactory results as the properties of linear
regression are normally distributed, which can generate negative or non-discrete values of accident rates
(Jovanis and Chang, 1986; Saccomano and Buyco, 1988; Miaou et al., 1991). As a result, Poisson and
negative binomial models were suggested to be used instead. Poisson models were suggested and used
by Jovanis, P. and Chang, H. (1986), Joshua, H. and Garber, N. (1990), Jones, B. et. Al (1991), Miaou,
S. and Lum, H. (1993), and Miaou (1994) for their consideration of random and sporadic data, in
addition to easy estimation of variables relationships, however results can be overestimated or
underestimated. When it comes to Poisson-gamma and negative binomial models, they have been
adopted by Maycock, G., and Hall, R. (1984), Hauer, E., Ng, J.C., and Lovell, J. (1988), Brüde, U.,
Larsson, J. (1993), Bonneson, J., McCoy, P. (1993), Miaou, S. (1994), Persaud, B. (1994), and Kumala,
R. (1995) for their handy over-dispersed data estimation, but they lack this property when dealing with
under-dispersed data. Generalized estimating regression can consider relationships changes over time
only if data is collected accordingly without any data loss or limitations related, as stated in studies done
in the early 2000s; such as Lord, D., and Persaud, B. (2000), and Lord, Mahlawat, (2009). Bivariate and
multivariate models can handle non-linear correlations by complex matrix operations and are usually
compared to generalized estimating processes in terms of functional forms. Some of the studies that
investigated bivariate and multivariate models were conducted by Miaou, S. and Lord, D. (2003), and
Ye, Pendyala, Washington, Konduri, and Oh (2009). Random-parameters models offer an accountancy
for heterogeneity of unnoticed data, but they require a long estimation process and a lot more difficult
of outcomes generalization, which was first adopted by Anastasopoulos, P., Mannering, F. (2009).
93%
34%
13%
Human Factore
Roadway Factore
Vehicle Factore
Sustainable Civil and Construction Engineering Conference
IOP Conf. Series: Earth and Environmental Science 357 (2019) 012040
IOP Publishing
doi:10.1088/1755-1315/357/1/012040
3
2.1. Geometric Design Elements Affecting Accidents Rates
A way to improve road safety is improving road geometric design to mitigate accidents occurrence and
severity on roadways. In order to improve the road design, it is crucial to evaluate and define the
relationship between road geometric design elements and road accidents.
2.1.1 Road Curves Radii. Road curve radius is a primary element of roads geometric design that is
associated with horizontal curve design and it is related to traffic accidents as the smaller the curve
radius, the higher the possibility of accidents to occur on roads. Based on the theory of vehicle steering,
vehicle transverse stability which includes the slippage and overturns determines the curve radius value
to be selected when designing road horizontal curves (Zhang Yingxue, 2009). Zhang computed the
radius mean and curvature degree for all categories used in his research and then he regressed them
against the mean of accident rates in every category, his findings supported some previous studies which
found and stated that the curve sharpness has a significant impact on the accident rates. Brude et al.
(1980) carried out a study on Swedish roads with a speed limit of 90 km/hr, the study aim was to develop
accidents prediction model to compare accident rates which occur due to horizontal curves to base
accident rate and it was concluded that the curve radius becomes significant factor when the curve radius
is about 0 m. Similarly, a study which was conducted by Department of Transportation, UK (1984)
pointed out that results showed a significant increase in accident rates when the curve radius reduces.
Other studies showed that the increase in accident rates becomes significant with roads which have curve
radius below 200 m. Simoson and Kerman (1982) noted that small radius curves can result in shorter
overall curve lengths and the overall impact on accident rates is not as awful as it might be expected,
however there is still an impact of curve radius on accident rates. Several previous studies performed by
Glennon et al. (1985), Glennon (1987), and Zegeer et al. (1991) showed that sharper curves are
associated with higher accident rates compared to milder curves which are associated with lower
accident rates. It was also found that horizontal curves have more accidents such as roll-over crashes,
opposite direction sideswipe crushes, etc. than other road sections (Ali Aram, 20). Different opinions
were found in terms of determining at which radius curve value the impact on road accident can reduce.
For instance, Oecd (1976) suggested that the critical curve radius to reduce the impact on accident rates
is 430 m and he concluded that the most accidents occur in curves are run-off accidents because of
driving dynamic aspects. On the other hand, Glennon et al. (1985) pointed out that greater curve radii
can increase the accident rates and costs and it was also concluded that road safety would decline when
the road curve radius is above 400 m. To conclude, all studies have pointed out that there is a significant
influence of curve radius on road accident rates and concluded that small radius curves are more
associated with higher accident rates and severity, however different views were encountered in
determining the curve radius value that can reduce the accident rates.
2.1.2 Road Gradients. Road gradient is one of the geometrical design elements of roads to be considered
in association with road accident rates as it has been agreed that it can have an impact on the accident
rates on roadways. According to some studies conducted, roads with steep gradients found to be related
to higher accident rates. Hedman (1990) quoted from a Swedish study done on road gradients and
accident rates that the results showed that road with 2.5 % grade can increase road accidents by around
% whereas roads with 4% grade can increase the accident rates by approximately 20 % compared to
near-horizontal roads. Glennon et al. (1987) investigated the results obtained for several studies
performed in US and concluded that roads which were designed with grade sections have a higher road
accident rates compared to roads which were designed with level sections. They also observed that
higher accident rates were found in steep gradient in comparison with mild gradient roads, it was also
noted that down-grade roads have more potential accident rates compared to up-grade roads. In
conclusion, the road gradient is significantly related to accident rates on roads as the increase in road
gradients results in an increase of road accident rates
Sustainable Civil and Construction Engineering Conference
IOP Conf. Series: Earth and Environmental Science 357 (2019) 012040
IOP Publishing
doi:10.1088/1755-1315/357/1/012040
4
2.1.3 Super elevation. Super elevation is defined as the traverse slope which is designed as higher on
the outer side and lower on the inner side. The super elevation is for the purpose of acting against a
centrifugal force that influences the vehicle’s running on curves. Also, to enhance the stability of
traveling vehicles and comfort of drivers. According to Zhang Yingxue 2009) the super elevation
traverse slope should be between 2.0 % to 3.0 %. Super elevation is a function of road horizontal
alignments, design speed, natural conditions and curve radius. He also mentioned that the use of proper
super elevation value can mitigate incidents which could contribute to accident severity. Some studies
have indicated that super elevation and horizontal alignment have an influence on traffic safety on roads.
A study conducted by Ali Aram (20) on two-lane highways concluded that sharper horizontal curves
drastically increase the rate of accidents occurrence; which is explained furthermore by the fact that
sudden changes in horizontal alignments along the road, declines drivers’ expectancy which, therefore,
accordingly, raises the probability of accidents occurrences. To reduce unexpectancy, few have stated
that consistent design speed can help maintain the attention of drivers to different road elements. Hence,
design speed criteria may differ for the same curve based on the maximized super elevation for each
criterion, as highlighted by AASHTO’s Washington, D.C.’s conference (2001). A correlation of both
actual and advised super elevation values is used by some two-way roadways design criteria,
particularly, a difference of more than 0.01. Moreover, these values can be obtained either from field or
2.1.4 Lanes Number. A considerable number of investigations indicate that lanes number is associated
with road accidents (Deo, Chimba, 2004). In a study of road geometrics, velocity and volume
relationships with road accidents; lane volume was found to affect accidents occurrences positively
(Garber, 2000). In a similar manner, a research finalized that urban highways lanes number increase
raises accidents rates (Abdel-Aty et al., 2000). Noland, R., and Oh.L. (2004) concluded similar outputs
of lane number positive influence on road accidents. Generally, various studies carried out suggest
corresponding findings ranging from two to six-lane roadways.
2.1.5 Lane Width. Generally speaking, freeway flow speeds are of higher values on lanes with ideal
widths; therefore, incidents are less possible, which is also stated by (HCM) especially for multilane
freeways (Jerry et al., 2009). While it might be rationally presumed that wider lanes reduce the effect of
incidents generating from driver mistakes, it can be argued that high operating speeds can oppose this
effect. The use of an optimum lane width value of typically 3.5 m to 3.6 m was suggested by most
researchers. A more reliable approach should be implemented such as consideration of more cross-
sectional elements related to traffic flow rates. Other studies highlighted that to some extent a wider lane
slightly increases road accidents (Hearne, 1976). On the other hand, another research implies a drastic
decline of road accidents with respect to wider lanes of the range 4.0 -7.0 meters or more than that
(Hedman K., 1990). However, several studies indicated that a much lower accidents occurrence is
associated lane widths from 3.4 m -3.7 m (Zegeer et al., 1981; Mclean, 1985; Zegeer and Council, 1993).
Similarly, NCHPT in the report 197, 1978 it was found that lower accident rates occur when the width
is 3.35 m or 3.65 m. In a study conducted by Hughes (1995) of rural characterized by low traffic flow
rates, it was concluded that lanes wider than 3.65 m reduce incidence occurrence. However, a lane wider
than 3.7 m does not necessarily imply safety provided other factors for instance, passing from
carriageway to another (Transportation Research Board, 1987). In addition, higher operating speeds of
wider lanes can result in more accidents. A wider lane than the minimum required results in a higher
speed which is a measure of driving comfort (Yager and Van Aerdo, 1983). It was noted that free flow
speed of highways of 3.3 m up to 3.8 m tend to be slower by 5.7 km/hr with respect to lane width
decrement of 1 m (O’Cinneide, 1995). Also, cross sections of 7.5 m or less showed a noticeable decline
in road accident rate as examined in a study by Lamm et al. (1999). Moreover, a research on anticipation
of road accidents occurring on non- intersection segments of four states in US concluded that wide cross
sections have a remarkable impact on accidents reduction of approximately 14% decline per 1 m
increase in width, while some states showed a higher percentage of 34 % (Council and Steward, 2000).
In terms of accident costs 3 m increase in width is linked to lower accident costs (Elvik, and Vaa, 2004).
Sustainable Civil and Construction Engineering Conference
IOP Conf. Series: Earth and Environmental Science 357 (2019) 012040
IOP Publishing
doi:10.1088/1755-1315/357/1/012040
5
As mentioned, widening carriageways does not necessarily provide safety improvement. To sum up, to
a certain extent wider lanes are attributed by lower accident occurrence, however, while this proves true,
safety is lower in wider cross sections (US Department of Transportation, 2007).
2.1.6 Sight Distance. Sight distance is one most critical factors that can contribute to road accidents. It
is known as the carriageway visible distance that can be seen and recognized on both vertical and
horizontal planes by vehicle drivers (Martin R., 2003). Inadequate sight distance can cause higher
accident rates as it is related reaction and decision to be taken by the driver to handle and deal with any
unforeseen circumstances or objects which the driver may encounter on the roadway. Sight distances
shall be taken into consideration during the design of the road horizontal and vertical alignments in order
to assure and provide sufficient and safe roadway design. The possibility of accidents occurring is higher
in some road sections with a poor visual distance due to a small radius of horizontal and vertical curves
and also with road sections where the sight distance of overtaking is not adequate to the drivers (Zhang,
2009). Sight distances involve stopping and passing sight distances. Although there is a lack of
information on the influence of stopping sight distance on the road safety, it is widely accepted that the
provision of shortstop sight distance can cause accidents on roads (O’Cinneide, 1995). Martin R. (2003)
published a book on highway engineering and he recommended different values of stopping and passing
sight distance at different design speed in order to ensure road safety, however he stated that there is no
evidence to prove that there is a relationship between accidents occurrence and passing sight distance.
Several studies were conducted in order to investigate the impact of different radii which correspond to
various sight distances. Transportation Research Board (1987) reported that roads which have shorter
sight distances because of crest vertical curves, the frequency of accidents occurrence is 52%. On the
other hand, Hall and Turner (1989) stated that insufficient stopping sight distance doesn’t necessarily
indicate that accidents will happen. Glennon et al. (1985) pointed out that the improvement of sight
distance on crest vertical curves are not significantly effective unless the traffic volume on the road is
high. Lamm et al. (1999) in their handbook on traffic safety and highway design mentioned that higher
road accidents frequency was observed on road sections with sight distances shorter than 0 m and there
is no further effective improvement on safety for sight distances above 150 m. Another study done by
Elvik and Vaa (2004) also concluded that having 200 m or more than 20m sight distances can increase
the possibility of accident risks on roads which means it can adversely affect the road safety.
2.1.7 Crest Curves. The minimum number of crest curves to be provided on a road section rely on the
stopping sight distances provision at all points. TRB (1987) in their report No. SR 214 developed an
equation to estimate the accidents rate on a road section which has one crest curve and it is approaches
of tangent. The report concluded that the geometrical design of vertical crest has not be proven to
significantly affect the frequency and severity on roadways. On the other hand, a search conducted by
Srinivasan (1982) showed that changes in vertical alignment can cause sight distances reduction at the
vertical crest curves which is observed to influence the accidents occurrence frequency.
2.1.8 Number and Density of Access. Access point density has been considered as significant dominant
of road accident rates on the highways worldwide. Some studies discussed the impact of access density
and number on the accident rates on roadways. For instance, Mouskos et al. (1999) have done a study
in New Jersey on the influence of access points on accident rates along multilane highways. They found
that about 30 % of accident rates on multilane highways occur in the mid-block sections and that can
explain the main reason which is the existence of access points. It was also found that about 25 % of
merging and diverging traffic at access points affect the traffic flow performance on multilane highways
which could result in accidents as well. Another study performed by Karlarftis et al. (2000) on rural
multilane highways showed that present of access points and medians have the most significant impact
on road accidents followed by the effect of pavement conditions. Several studies pointed out that the
road accident rates have a linear increase with the access density on highways, however some other
studies found that the increase in road accident rates due to access density is more than linear. For
instance, Gluck et al. (1999) developed a model to establish the relationship between access density and
Sustainable Civil and Construction Engineering Conference
IOP Conf. Series: Earth and Environmental Science 357 (2019) 012040
IOP Publishing
doi:10.1088/1755-1315/357/1/012040
6
accident rates and they suggested that the increase in the number of access points on road from -20
access points per mile would increase the accident rates by approximately 30 %. Similarly,
Papayannoulos et al. (2004) presented that the results obtained showed that the accident rates increase
as the number of access points increases. In their study, they have suggested that a road that has 60
access points per mile would result in triple accident rates compared to a road which has access points
per mile. In conclusion, based on the studies carried out to investigate the relationship between road
accidents and access density, all studies concluded that there is a significant influence of number and
density of access points in roads on the accident rates at that particular road
2,1.9 Shoulder Width. Shoulders are used as a free space to allow vehicles on roads to stop out of the
main road traffic flow whether for vehicle break-down incidences, emergencies or as a section of road
right of way (Deo Chimba, 2004). In case of control loss over the vehicle, shoulders are used as back up
to take back vehicle control (US Department of Transportation, 2007). More space provided by road
shoulders allow for high free flow speed because drivers may have a perception that when they lose
control over their vehicles there is a room for gaining control again. In addition, various road geometrical
design elements interact with one another (Jerry et al., 2009). Several studies agreed on shoulder width
impact on accident rates on roadways. Particularly, lane width is linked together with should width with
respect to road conditions (Deo Chimba, 2004). Studies carried out had different findings on the effect
of shoulder width on accident occurrence. Zegeer et al. (1981) stated that wider shoulders result in lower
accident rates and found that a decline up to about 20 % of incidence occurrence is attributed by shoulder
widths of 0.9 m – 2.7 m and they suggested that the optimum road shoulder width should be 1.5 m.
Turner et al. (1981) mentioned that shoulder width below 2.0m tend to decrease accidents occurrence in
comparison with shoulder widths of more than 2.0 m (Hedman, 1990). US Department of Transportation
(2007) reported that an increase of 0.3 m width, can reduce both incidents by 1.0 % to 3.0 % and
severities by 2.0 % to 4.0 %. According to Miaou (1996), an increase of shoulder width of 0.3 m can
also result in 8.8 % reduction of accident rates.
2.1.10 Median Width and Type. Road medians serve as a mean of directional traffic separation, can be
also used to prevent errant maneuvers, accommodate left-turn movement, and provide recovery area for
an emergency stop. In term of design, both median width and type are associated with median existence
(Jerry et al., 2009). When it comes to median type a study conducted by Hadi Et al.(1995) found that
raised curb, crossover resistance and stripped medians attribute to decline of road safety. Moreover,
physical narrow medians which have an impact if only vehicles crash into them are less effective than
wide medians. Also, land use along with median type highly influences accidents rate (Sawalha and
Sayed, 2001). It was concluded that on two-lane or multilane roadway with high design speeds, the road
safety is positively impacted by road medians as they can help avoid head-on crashes (Srinivasan, 1982).
According to The Road Directorate of Denmark (1981), Medians with barriers especially reduce
accident rates, however only to some extent of median width up to 3 m. On the other hand, other studies
suggested a decline of accident rates for medians wider than 12 meters (Hughes,1995).
2.1.11 Median Barriers. Studies examining barriers had different views regarding the impact of barriers
on accident rates (Nikiforos et al., 20). Hauer (2004) stated that the use of median may have a potential
influence on road safety. Council and Steward (2000) have concluded that safety is increased in the
existence of barriers. On the contrary, others suggested that barriers existence on the road is related to
left shoulder width (Fitzpatrick et al., 2008). Another group of researchers indicated a general increase
of accident rates with barriers presence, however a decline of the severity was observed (Elvik, 1995).
Moreover, the barriers present themselves as objects can potentially lead accidents on roads (Nikiforos
et al., 20). Particularly, concrete barriers have a higher possibility of accidents involvement (Elvik and
Vaa, 2004). Barriers can decrease accidents resulting from the errant manoeuvre, but also can increase
the number of accidents which are involved with medians as medians can be used as stopping space
(Nikiforos et al., 20). To conclude, barriers can decrease severe crossover accidents, thereby both type
and severity of accidents should be taken into consideration with and without barriers.
Sustainable Civil and Construction Engineering Conference
IOP Conf. Series: Earth and Environmental Science 357 (2019) 012040
IOP Publishing
doi:10.1088/1755-1315/357/1/012040
7
3.0 Discussion
One of the most widely adopted estimation methodologies is maximum likelihood estimation. Mainly,
maximum likelihood estimation is useful for its form functions, however, likelihood function requires
the definition of a more complex variable. The modelling of accidents rate has limitations; for instance,
sampling errors, over-dispersion, under-dispersion, time changing parameters and omitted variables. In
order to develop statistically accurate results, new methodologies have been embraced, such as random
parameter models. Moreover, other common limitations of data availability and variations of traffic
conditions are still affecting precision and generalization of findings. Therefore, accidents rate
prediction simulators along with detailed data approaches have been introduced, which is an indication
of the huge and promising possibilities. Past literature that investigated road geometric design elements
and road accidents relationship have shown that some of these elements are more significant than others.
For instance, lane width was found to be more crucial to crash rates than shoulder width, while this holds
true, certain types of crashes decreased by the increase in both lanes and shoulders widths, particularly,
opposite direction and run-off-road crashes, but to a certain width. Generally, accidents occurrence
declines with more lanes provided, on another notice, road curves result in nearly four times more crash
rates than road tangents. A shoulder width of up to 2.5 m have been found to maintain safety, but a wider
shoulder may raise crash rates. For homogenous traffic conditions, the horizontal curvature is more
effective on accidents rate than straight segments, specifically, at radii less than 200 m, which are
associated with an average operating speed of 90 km/h. Especially, graded horizontal curves have a
much higher impact on crash rates. In addition, studies have concluded that downgrades have a
significantly more effect on accidents rate compared to upgrades. Since steep grades are often short than
not, their influence on accidents occurrence is less effective than mild grades. On the other hand,
mountainous terrains cause 30% more accidents than level terrains. Medians have been found to reduce
head-on collisions, whereas, median barriers mitigate the severity of accidents. Studies interested in
sight distance have reached a conclusion that sight distance lower than 0 m positively impact crash rates,
however, barely affect safety.
4.0 Conclusion
There is a complexity encountered to understand the relationship between road geometric design
elements and road accidents rate because of the interrelations among the geometric design elements
themselves and also other accident-related factors, however it is obvious that some geometric design
elements such as short sight distance and small curve radius can significantly increase the road accidents
rate and severity. Moreover, a combination of certain geometric design elements can result in serious
accidents occurrence. It is also found that the assessment of current studies reliability is a challenge
because of variation in parameters definition and selection used in the studies, traffic volume and
composition, and lack of control on statistical data and models. In addition, the studies outcomes that
have been conducted in different countries may not necessarily be the same, applicable in other
countries, and should not be generalized because of difference in traffic flow conditions, driver
behaviour, environmental factors, actual road conditions, and road enforcement policies and practices.
However, there is quite wide international agreement on the impact of road geometric design elements
on road accidents. When it comes to statistical modelling of accident analysis and prediction, there is a
potential opportunity for road safety researchers to develop more reliable models through more research
work efforts and application of new statistical methodologies and approaches.
References
[1] Abdel-Aty, M.A., &Radwan, A.E. (2000). Modeling traffic accident occurrence and
Involvement. Accident Analysis and Prevention 34, P.P 633-642
[2] Ali Aram. (20). Effective safety factors on horizontal curves of two-lane highways. Journal
of Applied Sciences (22), Malaysia, P.P 2814-2822.
[3] American Association of State Highway and Transportation Officials (AASHTO). (2001). A
policy on Geometric Design of highways and Streets. Academic Press, Washington, D.C.
Sustainable Civil and Construction Engineering Conference
IOP Conf. Series: Earth and Environmental Science 357 (2019) 012040
IOP Publishing
doi:10.1088/1755-1315/357/1/012040
8
[4] Anastasopoulos, P., & Mannering, F. (2009). Anoteon modeling vehicle accident frequencies
with random-parameters count models. Accident Analysis and Prevention 41(1),153–159.
[5] Bonneson, J., & McCoy, P. (1993). Estimation of safety at two-way stop-controlled
intersections on rural roads. Transportation Research Record 1401, 83–89.
[6] Brüde, U., Larsson, J. (1993). Models for predicting accidents at junctions where pedestrians
and cyclists are involved. How well do they fit?. Accident Analysis and Prevention 25(5),499–509.
[7] Brude, U., J. Lassen, & H. Thulin. (1980). Influence of road alignment on traffic accidents.
VTI Meddelande, PP: 235.
[8] Council, F., & Steward, J.R. (2000). safety effects of the conversion of two-lane rural to
four-lane rural roadways based on cross-sectional models. Transportation Research
Board Annual Meeting.
[9] Deo Chimba. (2004). Evaluation of geometric and traffic characteristics affecting the safety of
six-lane divided roadways. M.Sc., The Florida State University College of Engineering.
[10] Department of Transportation. (1984). Highway link design advice Note TA 43/84, London.
[11] Department of Transportation. (2007). Safety Handbook for Secondary Roads. USA.
[12] Douglas, W., Joseoh, E., & Kelth, K.(2000). Operational and safety effects of highway
geometrics at the turn of the millennium and beyond .TRB, National Research Council,
Washington, D.C.
[13] Garber, N.J. (2000). The effect of speed, flow, and geometric characteristics on crash rates for
different types of Virginia highways". Virginia Transportation Council.
[14] Elvik, R. (1995). The safety value of guardrails and crash cushions. Accident Analysis and
Prevention, 27(4), P.P 523-550.
[15] Elvik, R., and Vaa,T.(2004), .The Handbook of Road Safety Measures . Amsterdam, Heidelberg
(u.a.), Elsevier.
[16] Fitzpatrick, K., Lord, D., & Park, B.(2008). Accident modification factors for medians on
freeways and multilane highways in Texas. TRB Annual Meeting CD-Rom.
[17] Glennon, J. (1985). Effect of alignment on highway safety, relationship between safety and key
highway features. SAR6, TRB Ltd., Washington, D.C., P.P 48-63.
[18] Gluck, J., Levinson H.S., & Stover, V.G. (1999). Impact of access management techniques,
NCHRP Report 420, National Cooperative Highway Research Program, Transportation Research
Board, Washington, D.C.
[19] Haddon, W. (1972). A logistic framework for categorizing highway safety phenomena and
activity. The Journal of Trauma, Vol. 12.Lippincon Williams and Wilkins, Washington, DC, pp. 193-
207.
[20] Hadi, M.A., Aruldhas, J., Chow, L.F., & Wattleworth, J.A.(1995). Estimating safety effects of
cross-section design for various highway types using negative binomial regression.
[21] Transportation Research Center, University of Florida
[22] Hall J.W., & Turner D.S. (1989). Stopping sight distance: can we see where we now stand.
Transportation Research Record 1208.
[23] Hauer, E. (2004). Statistical road safety modeling. Transportation Research Record 1897,81–87.
[24] Jones, B., Janssen, L., Mannering, F. (1991). Analysis of the frequency and duration of freeway
accidents in Seattle. Accident Analysis and Prevention 23(2), 239–255.
[25] Hearne, R. (1976). Selected geometric elements and accident densities on the network.
Environmental Research Unit, Dublin.
[26] Hedman. (1990). Road design and safety. Swedish Road and Traffic Research Institute, VTI
Report 35A, P.P 225-238, Sweden.
[27] Hedman, K.o. (1990). Road design and safety. Proceedings of Strategic Highway Research
Program and Traffic Safety on Two continents, Gothenburg, VTI Report 315 A.
Sustainable Civil and Construction Engineering Conference
IOP Conf. Series: Earth and Environmental Science 357 (2019) 012040
IOP Publishing
doi:10.1088/1755-1315/357/1/012040
9
[28] Hughes, W.E. (1995). Safety and human factors: worldwide review. International Symposium
on Highway Geometric Design Practices. Transportation Research Board, Boston.
[29] Jerry, Pigman, John, S., Wendel, R., & Dominique, L.(2009) .Impact of shoulder width and
median width on safety. NCHRP Report 633.Transportation Research Board of The National
Academies, Washington, D.C.
[30] Jovanis, P., & Chang,H. (1989). Disaggregate model of highway accident occurrence using
survival theory. Accident Analysis and Prevention 21(5),445–458.
[31] Jovanis, P., Aguero-Valverde, J., Wu, K.F., & Shankar, V. (2011). Analysis of naturalistic
driving event data: omitted-variable bias and multilevel modeling approaches. Transportation
Research Record 2236,49–57.
[32] Joshua, S., & Garber, N. (1990). Estimating truck accident rate and involvements using linear
and Poisson regression models. Transportation Planning and Technology15(1),41–58.
[33] Karlarftis, M.G., & Golias, I.(2000). Effect of road geometry and traffic volumes on rural
roadway accident rates ". Accident Analysis and prevention 34, P.P 357-365.
[34] Kumala, R. (1995). Safety at rural three- and four-arm junctions: development and applications
of accident prediction models. VTT Publications 233, Technical Research Centre of Finland, Espoo,
Finland.
[35] Lamm, R., Psarianos, B., & Mailaender, T.(1999). Highway Design and Traffic Safety
Engineering Handbook . New York, McGraw-Hill.
[36] Lord, D., & Mahlawat, M. (2009). Examining the application of aggregated and disaggregated
Poisson–gamma models subjected to low sample mean bias. Transportation Research Record 2136,1
[37] Lord, D., & Persaud, B. (2000). Accident prediction models with and without trend: application
of the generalized estimating equations procedure. Transportation Research Record 1717,2–8.
[38] Martin R. (2003). Highway Engineering. First Edition, Ireland.
[39] Maycock, G., & Hall, R. (1984). Accidents at 4-Arm Roundabouts. TRRL Laboratory Report
1120, Transportation and Road Research Laboratory, Crowthorne, U.K.
[40] Mclean, J.R. (1985). Accident-width relationship for single-carriageway rural roads.
Australian Road Research 15(4), P.P 271-275, Australia.
[41] Miaou, S. P. (1994). The relationship between truck accidents and geometric design of road
sections: Poisson versus negative binomial regressions. Accident Analysis and Prevention 26(4),471–
482.
[42] Miaou, S.P. (1996). Measuring the goodness-of-fit of accident prediction models. Federal
Highway Administration, Report No. FHWA-RD-96-040, USA.
[43] Miaou, S. P., & Lord, D. (2003). Modeling traffic crash-flow relationships for intersections:
dispersion parameter, functional form, and Bayes versus empirical Bayes. Transportation Research
Record 1840,31–40.
[44] Miaou, S. P., & Lum, H. (1993). Modeling vehicle accidents and highway geometric design
relationships. Accident Analysis and Prevention 25(6),689–709.
[45] Mouskos, K.C., Sun, W., % Qu.T. (1999). Impact of access driveways on accident rates at
multilane highways.National Center for Transportation and Industrial Productivity, New
[46] Jersey Institute of Technology.
[47] Nikiforos S., Dominique L., Jerry P., John S., & Wendel R. (20). Safety impacts of design
element trade-off for multilane rural highways. ASCE Journal of Transportation Engineering.
[48] Noland, R.B., & Oh.L.(2004). The effect of infrastructure and demographic change on traffic-
related fatalities and crashes: a case study of Illinois county-level data. Accident Analysis and
Prevention 36, P.P 525-532.
[49] O'Cinneide, D. (1995). The relationship between geometric design standards and safety.
University of College-Cork.
[50] Oecd. (1976). Hazardous road locations: identification and counter-measures. Organization for
Economic Cooperation and Development, P.P 8, Paris, France.
Sustainable Civil and Construction Engineering Conference
IOP Conf. Series: Earth and Environmental Science 357 (2019) 012040
IOP Publishing
doi:10.1088/1755-1315/357/1/012040
10
[51] Sawalha, Z., & Sayed, T. (2001). Evaluating safety of urban arterial roadways. Journal of
Transportation Engineering.
[52] Simoson D., & Kerman J.A. (1982). The research and development background to highway link
Design. Traffic Engineering and Control 23 (9), 1982.
[53] Srinivasan, S. (1982). Effect of roadway elements and environment on road safety. Institution
of Engineers, Vol.63.
[54] The Roads Directorate. (1981). Traffic engineering roads and path types. Catalogue of Types
for New Roads and Paths in Rural Areas. The Technical Committee on Road Standards, Copenhagen.
Transportation Research Board. (1987). designing safer roads. Special Report 214, TRB, Washington,
D.C, USA, P.P 319.
[55] Turner, D.S., Fambro, D.B., & Rogness, R.O. (1981). Effects of paved shoulders on accident
rates for rural texas highways. Transportation Research Record 819, P.P 30-37.
[56] Yager, M., &Van Aerde, R. (1983). Geometric and environmental effects on speeds of 2-lane
highways. Transportation Research, Vol. 17A, No.4.
[57] Ye, X., Pendyala, R., Washington, S., Konduri, K., & Oh, J. (2009). A simultaneous equations
model of crash frequency by collision type for rural intersections. Safety Science 47(3),443–452.
[58] Zegeer, & Council. (1993). Highway design, highway safety and human factors.Transportation
Research Circular 414, P.P 20-34, Transportation Research Board.Washington, D.C., USA.
[59] Zegeer, Deen, & Mayes. (1981). Effect of lane and shoulder widths on accident reduction on
rural two-lane roads, Transportation Research Board 806, P.P 33-34, Washington, D.C. USA.
[60] Zhang Yingxue. (2009) "Analysis of the relation between highway horizontal curve and traffic
safety. International Conference on Measuring Technology and Mechatronics Automation, China.
