Accident Analysis and Prevention 53 (2013) 2327

Contents lists available at SciVerse ScienceDirect

Accident Analysis and Prevention

journal homepage: www.elsevier.com/locate/aap

The effect of road and environmental characteristics on pedestrian hit-and-run

accidents in Ghana
Eric Nimako Aidoo , Richard Amoh-Gyimah, Williams Ackaah
CSIR-Building and Road Research Institute, Kumasi, Ghana

a r t i c l e

i n f o

Article history:
Received 15 August 2012
Received in revised form
12 December 2012
Accepted 14 December 2012
Keywords:
Pedestrian
Hit-and-run
Road accident
Logistic regression
Ghana

a b s t r a c t
The number of pedestrians who have died as a result of being hit by vehicles has increased in recent years,
in addition to vehicle passenger deaths. Many pedestrians who were involved in road trafc accident died
as a result of the driver leaving the pedestrian who was struck unattended at the scene of the accident.
This paper seeks to determine the effect of road and environmental characteristics on pedestrian hitand-run accidents in Ghana. Using pedestrian accident data extracted from the National Road Trafc
Accident Database at the Building and Road Research Institute (BRRI) of the Council for Scientic and
Industrial Research (CSIR), Ghana, a binary logit model was employed in the analysis. The results from the
estimated model indicate that fatal accidents, unclear weather, nighttime conditions, and straight and at
road sections without medians and junctions signicantly increase the likelihood that the vehicle driver
will leave the scene after hitting a pedestrian. Thus, integrating median separation and speed humps
into road design and construction and installing street lights will help to curb the problem of pedestrian
hit-and-run accidents in Ghana.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
Death and injury associated with road trafc accidents are
global phenomena that require urgent attention (Tay et al., 2009).
According to Peden et al. (2004), approximately 1.2 million people
worldwide are killed each year as a result of road trafc accidents.
In addition to increases in death of passengers involved in road
trafc accidents, pedestrian deaths that result from being knocked
down by vehicles have also increased in recent years. Pedestrian
accident has been found to be one of the single largest causes of
injury, disability and death in the developed world (Sullman et al.,
2011). Most accidents involving pedestrians are caused by the negligence of both the pedestrian and the driver. According to Hamed
(2000), both drivers and pedestrians in developing countries ignore
trafc regulations and thus take greater risks and exercise less
caution. Rosenbloom et al. (2004) described pedestrians as a contributing factor to trafc accidents, and they constitute a large
subgroup of those who are seriously and fatally injured. Pedestrians
continue to be the road user with the highest risk of death in trafc
in Ghana, and their actions predispose them to injury (DamsereDerry et al., 2010; Afukaar et al., 2011). The number of people who
are killed on the roads in Ghana is unacceptably high, and the death

Corresponding author at: CSIR-Building and Road Research Institute, Box UP 40,
Kumasi, Ghana. Tel.: +233 247581500; fax: +233 3220 60080.
E-mail address: en.aidoo@yahoo.com (E.N. Aidoo).
0001-4575/$ see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.aap.2012.12.021

of pedestrians is the leading cause of fatalities among urban road

users in Ghana (Ackaah and Adonteng, 2011; Damsere-Derry et al.,
2010). Ackaah and Adonteng (2011) showed that pedestrian collision alone resulted in approximately 51.7% of all fatal accidents in
Ghana during the period 20052007.
The death of many pedestrians who were involved in road trafc
accidents resulted from the driver leaving the pedestrian who was
struck at the scene of the accident. Pedestrians who did not die
immediately after the collision later died as a result of poor trauma
care (Mock et al., 1997; Peden et al., 2004). For example, most of
these accident victims may not have died if they had been rushed
to the hospital for medical treatment immediately following the
accident. Hit-and-run drivers also increase the victims exposure to
being struck again by a subsequent vehicle (MacLeaod et al., 2012).
Hit-and-run is described as the drivers decision to leave the scene
after hitting a pedestrian without reporting to the police.
Reducing the number of drivers who ee after hitting a
pedestrian will help to reduce deaths associated with pedestrian
accidents. According to Tay et al. (2009), one way to reduce hitand-run accidents is to increase the level of trafc enforcement.
These authors further argue that educating drivers may change
their attitudes and beliefs about hit-and-run accidents.
There are numerous diverse factors that contribute to hit-andrun accidents in general and in Ghana, in particular. This research
considers factors such as trafc control, road characteristics and
surface conditions, accident severity, light condition and day of
the week. Previous research has conrmed that these factors are

24

E.N. Aidoo et al. / Accident Analysis and Prevention 53 (2013) 2327

signicant inuences on the drivers decision to leave or stay on the

scene after a pedestrian accident. For instance, light condition, day
of the week and trafc condition have been found to be contributing factors in a drivers decision to leave the scene of the accident
after striking a pedestrian (Solnick and Hemenway, 1994; Johnson,
1997; Tay et al., 2008, 2009). Solnick and Hemenway (1995) and
MacLeaod et al. (2012) concluded in their research that pedestrian and environmental characteristics signicantly inuence the
drivers decision to participate in a hit-and-run by leaving the scene
of the accident. These researchers further found that drivers are less
likely to ee the scene of the collision if younger or older pedestrians are involved.
This paper aims to contribute to the existing literature on road
safety in Ghana by providing empirical results and identifying the
effect of road and environmental characteristics on pedestrian hitand-run accidents by drivers in Ghana. Identifying the effects of
road and environmental characteristics on pedestrian hit-and-run
accidents in Ghana is important for designing road safety policies
and ensuring the successful implementation of road safety programs that will reduce pedestrian mortality due to hit-and-run
accidents.
The remainder of the paper is organized as follows: Section 2
describes the methods employed in the study; Section 3 presents
the data, empirical analysis and results; and Section 4 provide concluding remarks.
2. Methodology
To explain the effect of the road, environmental and weather
conditions on the decision of the driver to leave or stay at the
scene of pedestrian accidents, a binary logistic regression model
was employed (Solnick and Hemenway, 1995; Tay et al., 2009).
This type of model was used because of the dichotomous nature
of the dependent variable (i.e., hit-and-run as opposed to not
hit-and-run). This type of model makes it possible to estimate
the probability of the driver leaving the accident scene based on
the independent variables incorporated into the regression model.
The independent variables incorporated into the regression model
can be categorical or quantitative in nature. The binary logistic
regression model is among the family of regression models in
the generalized linear model framework (Nelder and Wedderburn,
1972; Agresti, 2007). The model assumes a binomial distribution for
the binary dependent variable. A binary logistic regression model,
also known as a logit model, is dened in the following manner:
log

 P(Y = 1) 
1 P(Y = 1)

= log

  
1

= + 1 x1 + 2 x2 + + n xn ,

= log it()
(1)

where P(Y = 1) =  describes the probability of the driver leaving the scene after a pedestrian accident (hit-and-run), while [1
P(Y = 1)] represents the probability of the driver not leaving the
scene (Agresti, 2007). The logit model predicts the probability ()
of the driver leaving the scene of the accident. This probability falls
between 0 and 1 (0  1) for all possible independent variables.
In Eq. (1), is the intercept term in the model, i s (i = 1, 2, . . . , n)
are the regression coefcients for each covariate and xs are the set
of covariates/independent variables. The parameters in the model
can be estimated using a maximum likelihood approach. The estimated model can be evaluated by performing a likelihood ratio
test to determine the signicance of the covariates in the model.
Because the dependent variable is modeled using a log transformation, log it(), the interpretation of the estimated coefcient is
based on the exponential transformation of the estimated coefcient, which is commonly known as the odds ratio. For this study,
the odds ratio indicates the likelihood of the driver leaving the scene

Table 1
Descriptive statistics of pedestrian accidents (20042010).
Variable

No. of
accidents

No. of hitand-run

Year
2004
2005
2006
2007
2008
2009
2010

2704
3024
3208
3297
2982
3323
3040

236
217
257
222
243
259
234

8.7
7.2
8.0
6.7
8.1
7.8
7.7

Region
Ashanti
Brong Ahafo
Central
Eastern
Greater Accra
Northern
Upper East
Upper West
Volta
Western

3730
1149
2070
2929
8561
255
224
92
1140
1428

195
73
113
145
980
22
17
5
64
54

5.2
6.4
5.5
5.0
11.4
8.6
7.6
5.4
5.6
3.8

Road environment
Urban
Non-urban
Missing

13,940
7625
13

1197
471
0

8.6
6.2

Day of week
Weekdays
Weekend

14,900
6678

1124
544

7.5
8.1

Accident severity
Fatal
Non-fatal

5444
16,134

525
1143

9.6
7.1

20,088
1490

1492
176

7.4
11.8

Light condition
Day
Night no light
Night light on

14,612
4186
2767

934
451
281

6.4
10.8
10.2

Road description
Straight and at
Curve and inclined

19,631
1927

1568
94

8.0
4.9

Road separation
Median
No median

4238
17,312

343
1319

8.1
7.6

Road surface condition

Dry
Wet

21,386
143

1652
5

7.7
3.5

Road surface repair

Good
Rough

19,259
2269

1520
135

7.9
5.9

Location type
Not at junction
Junction

17,173
4360

1434
228

8.4
5.2

Trafc condition
None
Signage

12,055
9497

944
719

7.8
7.6

Weather condition
Clear
Others (rain, fog/
mist, etc.)

Hit-and-run (%)

Note: Percentages for missing observations are not shown in the table.

of the accident with respect to a change in the covariates being considered in the model. The data used in this research were extracted
from the National Road Trafc Accident Database at the Building
and Road Research Institute (BRRI) of the Council for Scientic
and Industrial Research (CSIR), Ghana. The database is compiled
from road trafc accident les from the Motor Trafc and Transport
Unit (MTTU) of the Ghana Police Service. The database at the BRRI
is subject to two shortfalls: non-reporting and under-recording.

E.N. Aidoo et al. / Accident Analysis and Prevention 53 (2013) 2327

In-depth information on the database and its shortfalls can be found

in Salifu and Ackaah (2011). The data used in this study represent
the period 20042010.
3. Empirical results and discussion
3.1. Descriptive statistics
Within the study period, a total of 21,578 pedestrian collisions
were recorded, with hit-and-run accidents amounting to 1668, or
7.7%, of the total. Ghanas pedestrian hit-and-run accident rate
(7.7%) is high when compared to countries such as Singapore, which
has a 1.8% rate (Tay et al., 2008), but it is low when compared with
8.1%, 17.9% and 18.1% reported for California (Tay et al., 2009), Calgary (Tay et al., 2010) and the United States (MacLeaod et al., 2012),
respectively.
Table 1 provides summary statistics of hit-and-run incidents
with respect to the independent variables considered. Table 1 further describes the frequency distribution of pedestrian collisions
as well as the number and percentage of hit-and-run accidents.
Annually, an average of approximately 238 drivers who struck a
pedestrian ed the scene. The table shows that hit-and-run accidents varied among the ten regions in Ghana with the highest
proportion (11.4%) occurred in the Greater Accra region and the
lowest proportion (3.8%) occurred in the Western region. Additionally, the proportion of hit-and-run accidents was higher in
urban areas than in non-urban areas (8.6% vs. 6.2%). In Ghana,
approximately 9.6% of fatal pedestrian accidents were caused by
hit-and-run drivers. The incidence of pedestrian hit-and-run accidents was higher during unclear weather conditions (such as rain
and fog), as well as during nighttime hours, especially in areas without street lights. Compared to roads with other characteristics, the
proportion of hit-and-run cases was lower on dry, straight, at and
good surface roads without median separation. The proportion of
hit-and-run accidents was not affected by the presence/absence of
signage on roads.
3.2. Results of logistic regression modeling
To determine the possible factors contributing to pedestrian
hit-and-run accidents, the binary logistic regression model was
tted to the set of available independent variables. The Akaike
Information Criterion (AIC) with backwards deletion technique
was used for the selection of the nal model for the data. The
nal model selected for the study omitted two non-signicant
independent variables (i.e., Year and Day of week). Thus, the nonsignicance of Day of week contradicted the studies of Solnick
and Hemenway (1995) and Tay et al. (2009), who found a higher
probability of hit-and-run on weekends than on week days. The
nal model was also evaluated for an over/under dispersion problem by performing a likelihood ratio test on the deviance residuals
(deviance = 11, 009, df = 21, 390). The result of the test was found
to be non-signicant at 1%, indicating that the model is satisfactory and does not have a problem of over/under dispersion. The
overall signicance (goodness-of-t) of the tted model was tested
using a likelihood ratio test. The test examines the signicant difference between the null model and the tted model. The results,
shown in the last column of Table 2, indicate that the tted model
is satisfactory.
Table 2 summarizes the maximum likelihood estimates of the
tted model. The coefcients for all of the variables were estimated
relative to the selected reference category.
Compared to the Western region (the reference group), the
number of pedestrian hit-and-run incidents was signicantly
higher in all other regions. The incidence was quite high in the

25

Greater Accra and Northern regions compared to all other regions in

Ghana, at approximately 279% and 100% higher, respectively, than
that of the Western region. The differences in hit-and-run accidents
among the regions in Ghana may be due to differences in ethnicity
and socio-cultural environment. There was also a higher likelihood
of hit-and-run incidents in urban areas compared to non-urban
areas, though this difference was found to be only marginally signicant (p-value = 0.0546). The incidence of hit-and-run accidents
in urban centers was approximately 13% higher than that of nonurban areas. This difference is because there is a higher possibility
that the driver may be identied by community members in the
non-urban areas and thus prevented the driver from leaving the
accident scene.
When the effect of accident severity on the likelihood of the
driver to leave the scene was examined, it was found that the pedestrian sustaining a fatal injury generally led to the driver leaving the
scene of the accident. The association between fatal accidents and
hit-and-run incidents might be due to the cost and legal implications involved. The driver is more likely to face severe punishment
when the injury sustained by the pedestrian is fatal, and s/he may
thus prefer to leave the scene.
Unclear weather conditions, such as rain, fog and mist, greatly
contributed to the incidents of hit-and-run accidents on Ghanaian
roads. The estimated parameters suggested that the likelihood of
hit-and-run as a result of these unclear weather conditions was
approximately 26% higher than for the clear weather condition.
Darkness also appeared to increase the likelihood of the driver
leaving the scene of the accident. The effect was approximately
68% and 38% higher during the night without light and during
the night with a light on, respectively, compared to the daytime.
Inadequate lighting may encourage hit-and-run behavior because
of the perceived lower probability of being identied (Tay et al.,
2008). Higher incidents of hit-and-run during unclear weather and
under these environmental conditions might be due to the drivers
inability to see the pedestrian. Additionally, for such conditions,
the driver is less likely to be identied by any witness, thereby
giving the driver the opportunity to ee the scene. This result is
consistent with the ndings of a similar study by Tay et al. (2009)
and MacLeaod et al. (2012). According to MacLeaod et al. (2012),
darkness makes it easier to protect the drivers identity even when
witnesses are available.
With respect to road description, roads that were curved and
inclined signicantly reduced the likelihood that the driver would
ee after hitting a pedestrian. The estimated value found was
approximately 25% lower than the hit-and-run incidents that
occurred on straight and at roads. Roads without median separation increased the likelihood of the driver leaving the scene of the
accident compared to the roads with median separation. Thus, the
trafc volume of separated roads, which is frequently higher, likely
prevents the driver from eeing from the scene after a accident. This
result is consistent with the ndings of Tay et al. (2009). According
to Tay et al. (2009), divided highways typically carry more trafc
volume than undivided highways and drivers are more likely to
be identied when they are involved in an accident on a divided
highway.
Regarding the road surface condition, the number of hit-and-run
incidents on roads with wet surfaces was not signicantly different
from dry surface roads. In addition, when the road surface repair
was rough (i.e., roads with potholes or untarred roads), the likelihood of the driver leaving the scene of the accident was signicantly
lower compared to roads with a good surface. This nding suggests
that the drivers ability to leave the scene of the accident at a high
speed may be inuenced by the rough nature of the road surface,
making it difcult for the driver to leave the scene. In this regard,
installation of speed humps in settlement areas with tarred roads
will reduce the speed of vehicles and likely prevent drivers from

26

E.N. Aidoo et al. / Accident Analysis and Prevention 53 (2013) 2327

Table 2
Parameter estimates for logistic regression model.
Variable

Coefcient

Standard error

Odds ratio

95% CI (OR)

Intercept

3.279

0.165

<0.001

Region
Ashanti
Brong Ahafo
Central
Eastern
Greater Accra
Northern
Upper East
Upper West
Volta
Western

0.369
0.474
0.457
0.288
1.333
0.694
0.585
0.219
0.370
Reference

0.162
0.188
0.173
0.166
0.150
0.266
0.290
0.485
0.191

0.022
0.012
0.008
0.084
<0.001
0.009
0.044
0.652
0.053

1.447
1.606
1.580
1.333
3.794
2.002
1.794
1.244
1.448

1.0541.986
1.1112.321
1.1272.215
0.9621.848
2.8255.095
1.1893.371
1.0163.168
0.4813.217
0.9952.106

Road environment
Urban
Non-urban

0.123
Reference

0.064

0.055

1.130

0.9981.281

Accident severity
Non-fatal
Fatal

0.508
Reference

0.057

<0.001

0.602

0.5380.673

Weather condition
Not clear
Clear

0.227
Reference

0.097

0.019

1.255

1.0391.516

Light condition
Night no light
Night light on
Day

0.521
0.322
Reference

0.066
0.078

<0.001
<0.001

1.683
1.380

1.4801.914
1.1851.608

Road description
Curve and inclined
Straight and at

0.282
Reference

0.114

0.013

0.754

0.6040.943

Road separation
No median
Median

0.269
Reference

0.071

<0.001

1.309

1.1391.504

Road surface condition

Wet
Dry

0.646
Reference

0.462

0.163

0.524

0.2111.297

Road surface repair

Rough
Good

0.202
Reference

0.097

0.038

0.818

0.6760.989

Location type
Junction
Not at junction

0.576
Reference

0.075

<0.001

0.562

0.4850.651

Trafc condition
Signage
None

0.038
Reference

0.055

0.483

0.962

0.8641.071

AIC
Pseudo R2
Overall goodness-of-t test (likelihood ratio test)

running away in the event of a accident. Similarly, drivers were

approximately 44% less likely to run away from a accident scene
when the accident occurred at junctions/intersections, such as
crossroads, staggered crossroads, junctions, or roundabouts, compared to straight sections of roadways.
With respect to trafc condition, the incidence of hit-and-run
on roads with signage, such as stop signs, pedestrian crossing, give
way, and trafc signals, was not signicantly different from roads
with no signs.
4. Conclusion
The number of pedestrians who have died as a result of being hit
by a vehicle has increased in recent years. Within the 7-year study
period, hit-and-run accidents accounted for 7.7% of pedestrian
collisions in Ghana. Comparing pedestrian hit-and-run accident situations in Ghana to that of other countries, the study found that

p-Value

11051.11
0.04
p < 0.001

Ghana has more hit-and-run incidents (7.7%) than Singapore (1.8%)

but fewer than California (8.1%), Calgary (17.9%) or the United States
(18.1%).
From the study, road and environmental factors that were found
to contribute to the number of hit-and-run incidents included lighting and weather condition, road description, median separation,
road surface condition and repair, location type of the accident and
trafc condition.
It is therefore recommended that median separation, speed
humps and road lighting be incorporated into road design and
construction. Incorporating these characteristics will signicantly
reduce pedestrian accidents and the likelihood of hit-and-run
incidents. Public awareness must also be a priority to address
pedestrian safety, particularly during the night and under unclear
weather conditions. Road safety education must be intensied in
the Greater Accra, Northern and Upper East regions, where the
incidence of hit-and-run is high.

E.N. Aidoo et al. / Accident Analysis and Prevention 53 (2013) 2327

Although the study was limited with respect to the effect of road
and environmental characteristics on pedestrian hit-and-run accidents, we believe that both drivers and pedestrian characteristics
are likely to inuence the drivers decision to leave the scene after
a pedestrian accident. For instance, the value of the pseudo R2 also
suggests the inclusion of other variables to the tted model. Hence,
it is therefore recommended that future research relating to pedestrian hit-and-run accidents should consider these characteristics to
improve policies regarding road safety.
Acknowledgements
We are grateful to Mr. Francis K. Afukaar (Chief Research Scientist) from the Building and Road Research Institute, Prof. Louis
Munyakazi from Kumasi Polytechnic, Dr. Moudud Md Alam from
Dalarna University and Prof. Johan Lyhagen from Uppsala University for their useful comments and suggestions. Any remaining
errors are our own responsibility. We are also thankful to the staff
of the Building and Road Research Institute.
References
Ackaah, W., Adonteng, D.O., 2011. Analysis of fatal road trafc crashes in Ghana.
International Journal of Injury Control and Safety Promotion 18, 2127.
Afukaar, F.K., Agyemang, W., Ackaah, W., Larbi, J., 2011. Road Trafc Crashes in
Ghana: Statistics 2010. Consultancy Service Report for National Road Safety
Commission of Ghana.
Agresti, A., 2007. An Introduction to Categorical Data Analysis, 2nd ed. John Willey
and Sons Inc., Hoboken, NJ.
Damsere-Derry, J., Ebel, B.E., Mock, C.N., Afukaar, F., Donkor, P., 2010. Pedestrians injury patterns in Ghana. Accident Analysis and Prevention 42, 1080
1088.
Hamed, M.M., 2000. Analysis of pedestrians behavior at pedestrian crossings. Safety
Science 38, 6382.

27

Johnson, C., 1997. Pedestrian fatalities on interstate highways: characteristics and

countermeasures. Transport Research Record 1578, 2329.
MacLeaod, K.E., Griswold, J.B., Lindsay, S.A., Ragland, A.D., 2012. Factors associated
with hit-and-run pedestrian fatalities and driver identication. Accident Analysis and Prevention 45, 366372.
Mock, C., Nii-Amon-Kotei, D., Maier, R.V., 1997. Low utilization of formal medical
services by injured persons in a developing nation: health service data underestimate the importance of trauma. Journal of Trauma 42, 504513.
Nelder, J.A., Wedderburn, R.W.M., 1972. Generalized linear models. Journal of Royal
Statistical Society A 140, 4877.
Peden, M., Scureld, R., Sleet, D., Mohan, D., Hyder, A.A., Mathers, C., 2004. World
Report on Road Trafc Injury Prevention. World Health Organization, Geneva.
Rosenbloom, T., Nemrodov, D., Barkan, H., 2004. Crossing at a red light: behaviour
of individuals and groups. Transportation Research Part F 7, 395404.
Salifu, M., Ackaah, W., 2011. Under-reporting of road trafc crash data
in Ghana. International Journal of Injury Control and Safety Promotion,
http://dx.doi.org/10.1080/17457300.2011.628752.
Solnick, S.J., Hemenway, D., 1994. Hit the bottle and run: the role of alcohol
in hit-and-run pedestrian fatalities. Journal of Studies on Alcohol 55 (6),
679684.
Solnick, S.J., Hemenway, D., 1995. The hit-and-run in fatal pedestrian accidents: victims, circumstances and drivers. Accident Analysis and Prevention 27, 643649.
Sullman, M.J.M., Gras, M.E., Font-Mayolas, S., Masferrer, L., Cunill, M., Planes, M.,
2011. The pedestrian behaviour of Spanish adolescents. Journal of Adolescence
34, 531539.
Tay, R., Barua, U., Kattan, L., 2009. Factors contributing to hit-and-run in fatal crashes.
Accident Analysis and Prevention 41, 227233.
Tay, R., Kattan, L., Sun, H., 2010. Logistic model of hit and run crashes
in Calgary. Canadian Journal of Transportation, North America. Available from: http://journals.hil.unb.ca/index.php/CJT/article/view/12412/17172
(accessed 03.05.12).
Tay, R., Rifaat, S.M., Chin, H.C., 2008. A logistic model of the effects of roadway,
environmental, vehicle, crash and driver characteristics on hit-and-run crashes.
Accident Analysis and Prevention 40, 13301336.

Further reading
Ackaah, W., Salifu, M., 2011. Crash prediction model for two-lane rural highways in
the Ashanti Region of Ghana. Journal of International Association of Trafc and
Safety Sciences 35, 3440.