Beruflich Dokumente
Kultur Dokumente
Volume 4, No 3, 2014
Copyright by the authors - Licensee IPA- Under Creative Commons license 3.0
Research article
274
ranging from around 9% in the Netherlands to 37% in Romania (NHTSA, 2012). In USA,
pedestrians represented 13% of all traffic fatalities and 3% of all traffic injuries during the
year 2010 (European Transport Safety Council, 2011). This could be explained by the fact
that, in most of the developing countries, there is a low auto ownership, so more journeys are
made on foot. Moreover, the behavior of pedestrians is particularly non-compliant and often
risk-taking, resulting in more likelihood of people being on the streets facing the risk of being
hit by vehicles. This increases the need to study and model pedestrian movement.
On most urban roads of Egypt, different types of vehicles move on the same road space
occupying any position on the road depending on availability of free space at a given instant
of time without complying to any lane discipline. Moreover, there is a complexity of
interactions between pedestrians, who have jaywalking behavior, and vehicular traffic which
increases mostly at uncontrolled mid-block and unsignalized intersections.
Based on the above, it is clear that pedestrian safety is a main issue for transport planners,
traffic engineers and policy makers. Hence, it is worth studying the road crossing behavior of
pedestrians in mixed traffic condition at uncontrolled mid-block location. This study is an
attempt in this direction.
1.1 Background
There have been several studies carried out to investigate the pedestrian behavior which is
influenced by different factors such as pedestrian perception, roadway, environmental
characteristics, etc.
Pedestrian crossing behavior is mainly governed by the gap acceptance theory. Each
pedestrian has a critical gap to cross the street. Many researches correlate the minimum gap
from the vehicle that is accepted by pedestrians who intend to cross streets at mid-block. This
parameter may be associated with traffic conditions and with vehicle and pedestrian
characteristics. In most of these researches (Oxley et al., 2005; Das et al., 2005) the distance
between the vehicles and the pedestrians appears to influence the minimum gap accepted by
pedestrians. In addition, an increase in traffic density leads to smaller accepted gaps. These
gaps are often described by means of probability distributions or are estimated by means of
linear regression modeling. Indicatively, it can be mentioned that the minimum accepted gap
has been estimated at two seconds and the mean accepted gap at eight seconds (Das et al.,
2005).
Behavioral analysis has revealed that pedestrians prefer rolling gap (pedestrian roll over the
small vehicular gaps) instead of waiting for larger gaps to cross the road (Brewer et al., 2006;
Kadali and Vedagiri, 2013).
Individuals judgment about when and where to cross the road are very complex and
normally represented by various factors such as comfort, convenience, ease of crossing, and
safety. If a pedestrian decides to walk, then the pedestrian cross the road somewhere on midblock and pedestrian behavior changes dynamically. Pedestrians continuously change their
actions with respect to environmental characteristics. Several researchers have attempted to
identify factors influencing pedestrian behavior including sudden decisions that affect
pedestrian walking characteristics such as choice to accelerate or decelerate walking speed,
stop or wait and where to cross a street (Ishaque and Noland, 2008).
275
Another issue is the decision of pedestrians to cross the road or not. Some studies has found
that it depends more on the distance between the vehicle and the pedestrian and not so much
on the related time gap (Chu et al., 2002; Sun et al., 2003). As a result, pedestrians may
choose inappropriate time gaps, because they are not able to estimate the actual speed of
incoming vehicles. However, other studies have found that it depends on the time gap (Chu et
al., 2002). Other parameters that affect crossing decisions include the presence of police
enforcement and the behavior of other pedestrians (Oxley et al., 2005; Lobjois and Cavallo,
2006). Discrete choice modeling is used by most researchers in order to estimate whether
pedestrians are going to cross a street at mid-block or not (Papadimitriou et al., 2009;
Lassarre et al., 2007).
Road crossing behavior with respect to gender has also been observed in various studies.
Males have a tendency to show more hazardous road crossing behavior than females due to
less waiting time (Khan et al., 1999; Tiwari et al., 2007).
Some studies have focused on detailed experiments to find out the effect of age on road
crossing decisions with effect of vehicle distance or speed of vehicle (Oxley et al., 1997;
Lobjois and Cavallo, 2007).
A recent study has focused on legal versus illegal pedestrian road crossing behavior at midblock location in China (Cherry et al., 2010). Few studies have identified pedestrian behavior
in mixed traffic streets and developed a micro-simulation model in order to find out the
fundamental characteristics as well as the conflicts of the pedestrian movement (Shahin,
2006).
A study in Beijing, investigated pedestrian behavior and traffic characteristics at unsignalized mid-block crosswalk. Authors have explained the pedestrian speed change
condition with pedestrian behavior (Jiangang et al., 2007).
A few studies have also been carried out at un-controlled intersections to explore the
interaction between motorists and pedestrians (Ibrahim et al., 2005).
Some studies have focused on the safety impacts of marked and unmarked crosswalks for
pedestrian and driver in rural and recreational areas (Mitman et al., 2010). Studies have also
explored the pedestrian road crossing behavior and safety issues at unmarked location
(Zhuang and Wu, 2011) and in narrow urban streets with mixed traffic condition (Kwon et
al., 1998).
Recently, a study in India has investigated, in more detail, the pedestrians' behavior at
uncontrolled mid-block location under mixed traffic condition (Kadali and Vedagiri, 2013).
The study developed models that include detailed behavioral parameters such as pedestrian
rolling gap, frequency of attempts, accepted lag or gap, number of observations while
crossing, frequency of disturbance while crossing, and driver yield behavior.
As a consequence, the results of these researches cannot be transferred and used in a national
setting like the one of Egypt, because the Egyptian road and transport network have different
characteristics and operational conditions. Not only are the road infrastructure and traffic
control often inadequate for pedestrians, but also the behavior of pedestrians is particularly
non-compliant and often risk-taking. This is partially reflected in the increased proportion of
road accidents involving pedestrians in Egypt.
276
In this context, the primary objective of this study is to describe the road crossing behavior of
pedestrian at uncontrolled mid-block location in Egypt. In particular, the study focuses on the
behavioral characteristics of pedestrian with the help of multiple linear regression (MLR) and
binary Logit (BL) models. The study tests the influence of the street crossing width variable
which has not been included in previous models.
2. Methodology
A field survey was carried out in nine different sites in Egypt. These sites were chosen to
have various crossing distances and adequate volume of pedestrians as well as vehicular
traffic to allow for collection of pragmatic behavioral data. The sites were chosen in three
cities: Port Said, Ismailia, and Cairo. The selected sites were uncontrolled (unmarked and no
right of way to the pedestrian) mid-block locations. Street crossing widths at the selected
locations ranged from 6 m to 11.25 m.
In this survey, pedestrian crossing decisions were videotaped in real traffic conditions. The
survey was conducted in October 2013 during working days in normal weather conditions.
The survey period was 30 minutes for each location. The video camera viewed a total of 3040 m length along longitudinal direction, out of this only 15 m was used for data collection
where the pedestrians are usually crossing the road. Ten JPEG files were obtained from each
second of video recording with the help of Snapshot Wizard software. An example of these
files is shown in Figure 1.
277
vehicle has just passed through vertical virtual line indicating the pedestrians crossing path.
Moreover, the waiting time of the pedestrian started when someone approached the pavement
until he set foot on the street. It is noted that these calculations included only the accepted
gaps and not the rejected ones. This survey also recorded the frequency of crossing attempt a
pedestrian makes due to the increase in waiting time at the curb or median. Pedestrian rolling
gaps were also extracted. Rolling gap means that pedestrian cross the road following zigzag
path to roll over the small vehicular gaps instead of waiting for larger gaps. This is a usual
behavior in developing counties. The speed of the incoming vehicle was measured at the
moment when the pedestrian just started to cross, and was calculated by dividing the distance
between the two virtual lines by the time the vehicle took considered to be constant during
the pedestrians crossing time. From each snapshot, individual characteristics were collected,
which include pedestrian gender, age, and whether he/she was accompanied by another
pedestrian. The collected variables are shown in Table 1.
Table 1: List of variables extracted from videotape
Variable
Street crossing width
Traffic gap
Vehicle speed
Waiting time
Frequency of attempt
Pedestrian speed
Distance
Gender
Age
Accompanied
Type of vehicle
Lane
Driver yielding
Crossing
Parking
Description (unit)
The crossed width between curb on one side and median or
opposite curb from the other side (m)
Time gap between two vehicles with reference to crosswalk
point (sec)
Speed of vehicle at crosswalk area (kph)
Time spent at the curb or median for suitable gap (sec)
Number of attempts a pedestrian makes to accept a vehicular
gap
Speed of pedestrian while crossing the road (m/sec)
The space between the vehicle and the pedestrian (m)
Female = 0, Male = 1
Elder (> 60 years) = 0
Middle (30 60 years) = 1
Young (< 30 years) = 2
Is the pedestrian accompanied by another pedestrian(s)?
No = 0
Yes = 1
Is the pedestrian carrying baggage?
No = 0
Yes = 1
Does the pedestrian roll over the available small gap?
No = 0
Yes = 1
Motorcycle = 0
Car, taxi, microbus = 1
Bus, truck = 2
Vehicle in nearside lane = 0
Vehicle in far-side lane = 1
Does the driver reduce speed or change his path when
pedestrian is crossing?
No = 0
Yes = 1
Does the pedestrian accept the gap?
No = 0
Yes = 1
Presence of illegally parked cars:
278
Variable
Description (unit)
No = 0
Yes = 1
(1)
Where:
Log-Gap: logarithm of accepted gaps;
VSPEED: vehicle speed (kph);
STW: street crossing width (m);
FATM: frequency of attempt a pedestrian makes;
RGAP: the pedestrian rolls over the available gap (= 1) or not (= 0); and
AGE: (Elder = 0) (Middle = 1) (Young = 2).
The model calibration was considered with 70% data and remaining data were used for
validation of the model. The calibrated R2 value was found as 0.795. The descriptive statistics
of MLR test, t, and p-values are summarized in Table 2. Reported t-values and pvalues are
the statistical test values of each independent variable. The comparison between observed
(remain 30% data) and predicted values showed the validity of calibrated model and a valid
R2 value was found 0.785. Also the critical gap was estimated by Raffs method and it
founded as 4.87 sec.
279
Table 2: Descriptive statistics results for the MLR gap acceptance model
Variable
Constant
VSPEED
STW
FATM
RGAP
AGE
i (Coefficient)
0.520
0.008
0.010
- 0.064
- 0.355
- 0.090
Standard error
0.028
0.005
0.022
0.013
0.009
0.016
t-value
20.602
6.208
3.355
- 2.111
- 3.344
- 2.800
p-value
0.001
0.000
0.019
0.008
0.000
0.022
(2)
(3)
Where:
U: the utility of choosing to cross the road at mid-block;
GAP : traffic gap (accepted or rejected) (sec);
RGAP: the pedestrian rolls over the available gap (= 1) or not (= 0);
FATM: frequency of attempt a pedestrian makes; and
VSPEED: vehicle speed (kph).
The descriptive statistics of BL Model test are summarized in Table 3. The significance of the
independent variable is considered with the effect of t-values and pvalues. The model
validation is carried out with success prediction table and the overall prediction accuracy was
found as 93.5%. Hence, the proposed model is strong enough to predict the gap acceptance
behavior at uncontrolled mid-block location.
Table 3: Descriptive statistics results for the binary Logit crossing choice model
i
(Coefficien
t)
- 3.981
2.701
6.423
3.200
- 0.301
Variable
Constant
GAP
RGAP
FATM
VSPEED
Standard
error
t-value
p-value
2.201
0.546
0.011
0.022
0.007
- 7.221
5.211
4.487
2.454
- 3.208
0.000
0.000
0.000
0.004
0.002
280
In order to show clearly to which extent each of the independent variables affects the
dependent variable, an analysis of elasticities (e) is carried out, as shown in Table 4.
Moreover, the relative effect (er), as a normalization of the estimated elasticities in relation to
the lowest elasticity, is calculated in order to compare the magnitude of effects of all
independent variables. The point elasticity (ei) was calculated for each individual in the
sample according to the following formula, whereas the overall elasticity (e) is calculated as
the average of (ei) in the sample:
ei = ( Yi / Xi)(Xi / Yi) = i (Xi / Yi)
(4)
Table 4: Elasticity of the accepted gap to the independent variables in the MLR model
Independent variable
Vehicle speed (VSPEED)
Pedestrian rolling gap (RGAP)
Frequency of attempt (FATM)
Street crossing width (STW)
AGE
Elasticity (e)
0.559
0.521
0.298
0.140
0.041
It is observed that if the variable age has an elasticity of 1, then the variable vehicle speed
has an elasticity of 13.6, that is it affects the gap acceptance 13.6 times more than the age.
The elasticity analysis shows that, among the different variables, vehicle speed and pedestrian
rolling gap are the most influencing variables on gap acceptance as they have the highest
elasticities. Frequency of attempts has medium effect while street crossing width and age
have the lowest effect.
Although type of vehicle is usually an important factor for accepting the gaps (Yannis et al.,
2010), in this study, it is observed that pedestrians accepts traffic gaps with respect to vehicle
speed, while it has the highest elasticity. It is true, because small vehicles may come with
higher speeds. The pedestrian may not accept the available gaps with small vehicle at higher
speeds and sometimes gaps with heavy vehicles may be accepted due to less speed. This
observation is strongly supported by recent studies in mixed traffic condition in developing
countries (Kadali and Vedagiri, 2013; Cherry et al., 2010). Therefore, speed of the vehicle
plays important roles in the MLR model.
It is clear that at mid-block locations, pedestrians rely mainly on rolling gap and accept traffic
gap size without waiting a long time after arriving at the curb. Table 5 shows the pedestrian
rolling gap behavior with available vehicular gap size. The mean accepted gap sizes in
seconds with and without rolling gap are 2.76 and 5.22, respectively, which is intemperate
difference. If pedestrians choose rolling gap they are more likely to accept the minimum gap
sizes. This proves that the behavior of pedestrians in Egypt is particularly non-compliant and
often risk-taking, which is usual in many developing countries. This observation is strongly
supported by recent study in mixed traffic condition in India (Kadali and Vedagiri, 2013).
Table 5: Minimum accepted gap size under pedestrian rolling gap behavior
Rolling gap condition
With rolling gap
Without rolling gap
Minimum
1.5
3.1
Mean
2.76
5.22
281
Street width and pedestrian age are also important variables in MLR model for reducing gap
size, but they have insignificant effect in crossing choice (BL model).
It is observed from Table 2 that pedestrians accept traffic gaps with respect to the crossed
width between the curbs (STW). The pedestrians may accept smaller gaps with short crossing
distance. This is attributed to the fact that they reach the opposite curb in shorter safe time.
Pedestrian age is statistically significant for minimum gap size and there is a significant
difference between elders and young pedestrian age groups, which can be observed in Table
6. The mean accepted gap sizes in seconds for elders, middle, and young age groups were
5.85, 3.38, and 3.37 respectively. This indicates that the pedestrian chooses small gap sizes
with decrease in age, but there is not much difference between middle and young age groups.
This observation is strongly supported by recent study in mixed traffic condition in
developing countries (Kadali and Vedagiri, 2013).
Table 6: Minimum accepted gap size for different age groups
Age group
Elders
Middle
Young
Minimum
2.7
1.5
1.1
Mean
5.85
3.38
3.37
In the BL model for mid-block crossing choice, only four variables; gap size, rolling gap,
frequency of attempts, and vehicular speed were significant and included in the model.
Moreover, there is a probability of increasing pedestrian gap acceptance with the increase of
the gap size, rolling gap, and frequency of attempt, whereas, it reduces with the increase in
vehicle speed. It is interesting to note that none of the pedestrians' individual characteristics
were found to be significant in the crossing choice model. This is attributed to the fact that
most of survey participants may have a strong familiarity with the survey site, as this is
located in a very central area, resulting in less uncertainty in the decisions of those groups of
pedestrians that are often associated with particular behaviors.
The elasticity analysis for the BL model is presented in Table 7. The pedestrian rolling gap
has the greatest effect on pedestrians decision to cross the street or not. As expected, it was
found that if pedestrians choose rolling gap, they more likely decide to cross the street. The
variable with the second greater effect is the traffic gap size. As expected, the higher the
available gaps, the easier the crossing. Moreover, frequency of attempt has the lowest effect
on crossing decision.
Table 7: Parameter elasticities for the crossing choice (BL) model
Independent variable
Pedestrian rolling gap (RGAP)
Traffic gap (GAP)
Vehicle speed (VSPEED)
Frequency of attempt (FATM)
Elasticity (e)
0.451
0.332
0.113
0.013
5. Conclusions
A field survey was carried out at nine uncontrolled mid-block locations in different streets in
three Egyptian cities. The purpose was to investigate pedestrian road crossing behavior under
282
mixed traffic condition in urban areas. These behavioral characteristics are very useful to
control pedestrian jaywalking behavior and for improving pedestrian safety.
A lognormal regression model was developed in order to examine the effect of various
parameters on the size of traffic gaps accepted by pedestrians. It was found that the accepted
gaps depend on the speed of incoming vehicle, the pedestrian rolling gap, the frequency of
attempts before crossing, the crossing width, and the age of the pedestrians.
A binary Logit model was also developed in order to examine the effects of various
parameters on the decision of pedestrians to cross the street or not. The results suggest that
pedestrians decision to cross the street depends on the size of traffic gap, the vehicle speed,
the pedestrian rolling gap, and the frequency of attempts before crossing.
The results of this research show that pedestrians rely mainly on rolling gap and accept small
traffic gap size without much waiting at the curb, which proves that the behavior of
pedestrians in Egypt is particularly non-compliant and often risk-taking, which is usual in
many developing countries.
Pedestrians' individual characteristics were found insignificant in crossing choice; only
pedestrian's age and frequency of attempts were found to affect gap acceptance.
It was found that pedestrians accept vehicular gaps with respect to vehicle speed rather than
vehicle type. This can be justified by the fact that small vehicles may come with higher
speeds and heavy vehicles may come with less speed.
One of the new variables in the gap acceptance model is the crossed width between the curbs.
It was observed that pedestrians may accept smaller gaps with short crossing distance. This is
attributed to the fact that they reach the opposite curb in shorter safe time. Nevertheless, this
variable was not significant in the crossing choice model.
6. References
1. Brewer M. A., Fitzpatrick K., Whitacre J. A., and Lord D., (2006), Exploration of
pedestrian gap-acceptance behavior at selected locations, Transportation research
record 1982, Journal of the transportation research board, Washington D.C., pp 132140.
2. Cherry C., Donlon B., Yan X., Moore S. E., and Xiong J., (2010), Illegal mid-block
pedestrian crossings in China: gap acceptance, conflict and crossing path analysis,
Journal of injury control and safety promotion, 19(4), pp 320-330.
3. Chu X., Gittenplan M., and Baltes M., (2002), Why people cross where they do - The
role of the street environment, In the proceedings of the TRB 81st annual meeting,
Transportation research board, Washington.
4. Das S., Manski C.F., and Manuszak M., (2005), Walk or wait? An empirical analysis
of street crossing decisions, Journal of applied econometrics, 20(4), pp 445-577.
5. European Transport Safety Council (2011), Unprotected Road Users: a Key Concern
of Road Safety, Road safety performance index (Flash 19) report, ARRB Group
Limited, Vermont South, Victoria.
283
6. Ibrahim J., Day H., Hirshon J., and El-Setouhy M., (2012), Road risk perception and
pedestrian injuries among students at Ain Shams University, Cairo, Egypt, J. Inj.
Violence Res., 4(2), pp 65-72.
7. Ibrahim N. I., Kidwai F. A., and Karim M. R., (2005), Motorists and pedestrian
interaction at un-signalised pedestrian crossing, Proceedings of the Eastern Asia
society for transportation studies, 5, pp 120-125.
8. Ishaque M. M. and Noland R. B., (2008), Behavioral issues in pedestrian speed
choice and street crossing behavior: A review, Transport reviews: A transnational
trans-disciplinary, 28(1), pp 61-85.
9. Jiangang S., Yanyan C., Futian R., and Jian R., (2007), Research on pedestrian
behaviors and traffic characteristics at unsignalized midblock crosswalk: a case study
in Beijing, Transportation Research Record 2038, Journal of transportation research
board, Washington D.C., pp 23-33.
10. JTI (2011). Traffic Accidents in Jordan 2010, Jordan traffic institute, Amman,
Jordan.
11. Kadali B. R. and Vedagiri P., (2013), Modelling pedestrian road crossing behavior
under mixed traffic condition, European transport, 3 (55), ISSN 1825-3997.
12. Khan F. M., Jawaid M., Chotani H., and Luby S., (1999), Pedestrian environment
and behavior in Karachi, Pakistan, Accident analysis and prevention, 31(4), pp 335
339.
13. Kwon Y., Morichi S., and Yai A., (1998), Analysis of pedestrian behavior and
planning guidelines with mixed traffic for narrow urban streets, Transportation
Research Record 1636, Journal of the transportation research board, Washington
D.C., pp 116-123.
14. Lassarre S., Papadimitriou E., Yannis G., and Golias J., (2007), Measuring accident
risk exposure for pedestrians in different micro-environments, Accident analysis and
prevention 39, pp 1226-1238.
15. Lobjois R. and Cavallo V., (2006), Age-related differences in street crossingdesicions: The effects of vehicle speed and time constraints on gap selection in an
estimation task, Accident analysis prevention 39 (5), pp 934-943.
16. Lobjois R. and Cavallo V., (2007), Age-related differences in street-crossing
decisions: The effects of vehicle speed and time constraints on gap selection in an
estimation task, Accident analysis & prevention, 39(5), pp 934-943.
17. Mitman M. F., Cooper D., and Dubose B., (2010), Driver and pedestrian behavior at
uncontrolled crosswalks in tahoe basin recreation area of California, Transportation
research record 2198, Journal of the Transportation research board, Washington
D.C., pp 23-31.
18. Mohan D., Tsimhoni O., Sivak M., and Flannagan M. J., (2009), Road safety in
India: Challenges and opportunities, University of Michigan, USA.
284
19. NHTSA (2012), Traffic Safety Facts 2010 Data: Pedestrians, Report Number: DOTHS-811-625, National highway traffic safety administration, Washington DC, USA.
20. Oxley J., Fildes B., Ihsen E., Charlton J., and Day R., (1997), Differences in traffic
judgments between young and old adult pedestrians, Accident Analysis &
Prevention, 29(6), pp 839-847.
21. Oxley J., Fildes B., Ihsen E., Charlton J., and Days R., (2005), Crossing roads safely:
An experimental study of age differences in gap selection by pedestrians, Accident
analysis and prevention 37, pp 962-971.
22. Papadimitriou E., Yannis G., and Golias J., (2009), A critical assessment of
pedestrian behavior models, Transportation research part F 12, pp 242-255.
23. Shahin M. M., (2006), Pedestrian behavior with mixed traffic in developing
countries, Traffic engineering and control, 47(8) pp 303-309.
24. Sun D., Ukkusuri S. K., Benekohal R. F., and Waller S. T., (2003), Modeling of
motorist-pedestrian interaction at uncontrolled mid-block crosswalks, 82nd TRB
Annual Meeting, Transportation research board, National research council,
Washington D.C.
25. Tiwari G., Bangdiwala S., Saraswat A., and Gaurav S., (2007), Survival analysis:
pedestrian risk exposure at signalized intersections, Transportation research part F:
Traffic psychology and behavior, 10 (2), pp 77-89.
26. Yannis G., Papadimitriou E., and Theofilatos A., (2010), Pedestrian gap acceptance
for mid-block street crossing, World transport research -proceedings from the 12th
World Conference on Transport Research, pp 1-11.
27. Zhuang X., and Wu C., (2011), Pedestrians crossing behaviors and safety at
unmarked roadway in China, Accident analysis & prevention, 43(6), pp 1927-1936.
285