Sie sind auf Seite 1von 6

Accident Analysis and Prevention 43 (2011) 18461851

Contents lists available at ScienceDirect

Accident Analysis and Prevention


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

Distance between speed humps and pedestrian crossings: Does it matter?


Charlotta Johansson , Peter Rosander, Lars Leden
Lule University of Technology, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: Speed humps are a common physical measure installed at pedestrian crossings to reduce vehicle speeds
Received 3 April 2009 therefore improve the safety and mobility of pedestrians at the crossing.
Received in revised form 13 April 2011 The aim of this study was to determine whether variations in distance between speed humps and
Accepted 16 April 2011
pedestrian crossings contribute differently to the safety and mobility of pedestrians and cyclists, espe-
cially children and the elderly, and if so, how. Three sites in Sweden were studied, where vehicle speed
Keywords:
measurements and video lming at the site resulted in manually coded, road user behaviour of 1972
Trafc safety
pedestrians and cyclists.
Pedestrians
Bicyclists
Road user behaviour at three test sites and two comparison sites equipped with speed cushion at
Speed humps distances of about 5 m and 10 m from the pedestrian crossing, i.e. about one or two car lengths, were
Physical speed reducing measures studied. As vehicle speeds were somewhat lower at the pedestrian crossing when the distance between
the speed cushion to the pedestrian crossing was greater, and there were positive aspects regarding the
mobility of the pedestrians and cyclists, a greater distance of about 10 m or two car lengths between the
hump and the pedestrian crossing is suggested. The present study only covers speed cushions, but the
same distance is also regarded as important when installing other types of physical measures to reduce
vehicle speed.
2011 Elsevier Ltd. All rights reserved.

1. Introduction show a shorter distance, see e.g. Leden et al. (1982) and Vgverket
(2004).
Actual vehicle speed should be a maximum of 30 km/h where The aim of this study is to determine how various distances
there is a risk of collision between vehicles and unprotected road between speed cushions and crossings affect safety and mobility
users. At speeds of 50 km/h, the probability of pedestrian death for pedestrians and cyclists, especially children but also the elderly.
is 58 times greater than at 30 km/h (Ashton, 1982; Teichgrber, Road user behaviour at three test-sites and two comparison sites
1983; Waltz et al., 1983). Speed cushions are a common type equipped with speed cushion at distances of about 5 m and 10 m
of physical measure used to increase safety and the mobility for from the pedestrian crossing, i.e. about one or two car lengths,
pedestrians and cyclists at crossings and cycle routes in built-up were studied. Data for these distances was available from previous
areas. These measures are recorded as being very effective (Elvik projects and a short exploratory analysis suggested that pedestri-
and Vaa, 2004). Speed proles for trafc calmed roads as a func- ans were more often given way to by drivers when the distance was
tion of the local speed, the type and design of the measure, see e.g. longer, but the amount of data were not sufcient to draw rm con-
Leden et al. (2008), and the distance between measures, see e.g. clusions. Therefore it was decided to collect more data. The study
Barbosa et al. (2000) has been extensively researched. Also drivers only covers speed cushions, but it is likely that the same distance is
and pedestrians behaviour at pedestrian crossings has been exten- suitable for the location of other types of speed reducing devices.
sively researched; see e.g. Vrhelyi (1998) and Johansson (2004).
However, the inuence of the distance between a speed hump and
2. Method
a pedestrian crossing has not been in focus in any previous study
as far as we have found. And a search on speed humps and dis-
The following methods of data capture were used: measurement
tance in Science Direct gave no relevant search results at all. At
of vehicle speed, and video recording of road user behaviour.
least Swedish guidelines do not discuss the distance between the
A coding system to describe school childrens travel to and from
speed hump and the crossing explicitly, but illustrations seem to
school proposed by (vstedal and Ryeng, 1999) was further devel-
oped and used to describe also other road users behaviour. The
coding describes the mobility and security of the pedestrians and
Corresponding author. Tel.: +46 920 491867. bicyclists, e.g. if any drivers yield to them, stopping at the kerb, and
E-mail address: Charlotta.M.Johansson@ltu.se (C. Johansson). if they (especially children) start to run when crossing the second

0001-4575/$ see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aap.2011.04.020
C. Johansson et al. / Accident Analysis and Prevention 43 (2011) 18461851 1847

Table 1
Study sites and overall data.

Data Total number hours Total number observations

Regementsgatan/Dragonstigen, Malm 20012006 36 818


Tessins Vg/Henrik Wranrs Gata, Malm 2006 9.5 437
Hultagatan, Bors 2002 16.5 717
Total 62 1972

lane. Also, safety is described through these types of parameters age groups was encoded if they were in the company of the above
while lack of stopping at the kerb and looking around before cross- groups. The parameters studied are given below.
ing the street are examples of errors made by especially children in (a) sex (k) pace before crossing
so called dash-out accidents. These were also the most important (b) age (l) pace on crossing rst lane
parameters for describing safety according to an expert question- (c) mode of transport (m) pace on crossing second lane
(d) number of persons in group (n) pace after crossing
naire (Johansson and Leden, 2007) and those parameters were
(e) sex of oldest in group (o) trafc situation, related to the rst
chosen for analysis. The experts rankings of the most important driver
parameters were in rank: (f) age of oldest in group (p) whether rst or following drivers
from left give way
(g) whether they stop at edge of the (q) whether rst or following drivers
(a) speed of vehicle, pavement, kerb from right give way
(b) speed of the pedestrian or bicyclist, (h) whether they stop on the refuge (r) long or short distance to speed
cushion
(c) at what distances evasive actions are taken,
(i) whether they cross at the crossing (s) driver behaviour in giving way
(d) if the pedestrian or cyclist look around before crossing the (j) whether they cross diagonally or (t) comments
street, and straight
(e) whether or not she or he stops at the kerb before crossing the Two observers coded all the data, and discussed unclear cases
street. together to obtain a consistent coding. By comparing 1.5 h of video
material comprising 47 cases, the two observers had 83% agree-
2.1. Measurement of vehicle speed ment in behaviour coding. When the most difcult parameter to
encode, pedestrian and cyclist head movement, was eliminated,
Radar equipment was used in the beginning of the project to agreement was 89% (Johansson, 2001). During the project the
measure speeds. Laser equipment which is more accurate than observers continued to discuss unclear cases to obtain uniformity.
radar, were used for the majority of measurements. Only the speeds Differences in results depending on long or short distances
of free vehicles were measured, i.e. the vehicle drivers were not between speed cushion and pedestrian crossing, and speed cush-
interacting with other road users. Radar and laser measurements ion and cycle crossing, were analysed using the 2 -test, arranged
of free vehicle speed were taken at the pedestrian crossings, i.e. the- into a frequency or contingency table (Siegel and Castellan, 1988).
oretical collision site. Speeds were also measured 12 m before the The rows in the table represents short and long distance to the
pedestrian crossing, the last point at which drivers need to react to speed cushion and each column represents a category of the mea-
be able to stop before reaching the pedestrian crossing if they travel sured variable, i.e. given way by some driver (yes or no). Signicant
at speeds around 38 km/h, have a perception reaction time of 0.5 s results (p < 0.05) are marked.
and a braking deceleration of 8 m/s2 . With this deceleration rate, a
driver who starts braking at the 12 m point could stop in time as 3. Description of sites and data capture
long as their speed is below 50 km/h. On the other hand, an inatten-
tive driver would typically have a perception-reaction time much Studies were made at three sites in Sweden, the Rege-
longer than 0.5 s, and speeds would need to be below 30 km/h at mentsgatan/Dragonstigen and Tessins Vg/Henrik Wranrs Gata
this spot to provide reasonable safety for pedestrians facing such crossings in Malm, and a pedestrian and cycle crossing on Hul-
drivers. tagatan in Bors, where speed cushions have been installed earlier.
The aim of the measures was to reduce vehicle speed at the con-
ict points with unprotected road users. At two of the sites there
2.2. Video recording of road user behaviour
were cycle crossings which meant that the distance to the pedes-
trian crossing was differently depending on the direction of the
At each site, video recordings were made from different angles
oncoming vehicular trafc. At the third site the speed cushions
in order to capture the behaviour and encounters of various road
were installed at different distances from the pedestrian crossings
users. Recordings of the crossings were made different occasions
as suggested by the three authors to allow for studying the effect
during 2 days, with two cameras giving a general view of the cross-
of the distance between speed cushions and pedestrian crossings.
ing and close-ups of pedestrians and cyclists. The sites were videoed
All three sites are equipped with central refuges.
from 07:30 to 09:00 and for three successive hours in the interval
The sites were studied over different periods, see Table 1. A total
13:0016:30, in total 4.5 h per day. Data entry from video record-
of 1972 pedestrian and cyclist crossings were studied from a total
ings was carried out manually in Excel in accordance with the
of 62 h of lm.
parameters given below. The parameters can be placed in three
groups, (1) description of the pedestrian or the cyclist and their
behaviour, (2) description of the vehicle and driver behaviour, and 3.1. Regementsgatan/Dragonstigen, Malm
(3) description of the interaction between them.
No conicts were registered so it was not possible to use the There are two schools in the vicinity of the two study sites
parameter at what distances evasive actions are taken. Codings in Malm, which means that many children pass Regementsgatan
were stratied so that behaviour for children (012 years), young and Tessins Vg on their way to or from the school or during their
people (1319), elderly (over 64) and functionally disabled irre- free time. There are also many elderly and disabled people in the
spective of age, was encoded. The behaviour of adults in the 2064 area, and there is bus trafc as well as some other heavy trafc
1848 C. Johansson et al. / Accident Analysis and Prevention 43 (2011) 18461851

Fig. 1. Dimensions of the DragonstigenRegementsgatan crossing looking west. The


shorter distance is 5.07 m to the left in the gure, and the longer distance to the right
is 8.78 m.

such as goods vehicles. The crossing at Dragonstigen is a pedes-


trian and cycle crossing on a section. There are speed cushions at
either 5.1 m or 8.8 m before the pedestrian crossing, and either 5.2
or 8.7 m before the cycle crossing, see Fig. 1. The height of the speed
cushions is about 100 mm. White cobblestones are used to indicate Fig. 2. Plan of the Tessins Vg/Henrik Wranrs Gata crossing after reconstruction.
The shorter distance is 4.2 m to the left in the gure, and the longer distance to the
the pedestrian crossings. The speed limit is 50 km/h. Vehicle ow at right is 10.2 m.
the site was around 9000 vehicles a day according to measurements
2005 made by Malm municipality.
4.1. Vehicle speeds
3.2. Tessins Vg/Henrik Wranrs gata, Malm
The two sites in Malm with different distances between speed
Tessins Vg goes through Ribersborg in the western part of cen- cushion and pedestrian crossing had similar motor vehicle speeds
tral Malm. The distance between the pedestrian crossing and the at the pedestrian crossing in both directions measured on free vehi-
speed cushion is 4.2 m for trafc travelling north, and 10.2 m for cles. What was unexpected was that, when the distance to the
vehicles travelling south-west. The distance between the pedes- pedestrian crossing was longer, speed was somewhat lower (not
trian crossings on each side of the crossing is approximately 19 m. statistical signicant) at the pedestrian crossing, just over 22 km/h
Crossings and speed cushions are indicated with painted mark- at longer distances and just over 23 km/h at shorter distances, see
ings, and the height of the speed cushions is about 100 mm. The Table 2. A possible explanation could be that drivers were more
posted speed limit through the crossing is 30 km/h between 07:00 cautious, tentative and more aware of the approaching pedestrian
and 17:00 on weekdays, but not during the three summer months crossing when the speed cushion was situated at a longer dis-
(16/614/8) when a 50 km/h limit applies instead. Tessins Vg had tance from the pedestrian crossing as drivers then got more time
a vehicle ow of 4000 vehicles per day in 2005 (Fig. 2). to focus on approaching pedestrians and cyclists after passing the
speed cushion. With the speed cushion at a shorter distance from
the pedestrian crossing drivers might be more focused on pass-
3.3. Hultagatan, Bors
ing the speed cushion, and therefore adjust their speed less to the

The Hulta centre with junior and middle schools and businesses
lies south of the area studied. Hultagatan had a daily vehicle ow
of 5000 vehicles in 2001. Posted speed limit is 50 km/h but the rec-
ommended speed is 30 km/h. The distance between the pedestrian
crossing and the speed cushion is 3 m for trafc travelling east, and
8 m for vehicles travelling west, see Fig. 3. The height of the speed
cushions is 70 mm.

4. Effect of distances

The analyses concern the effect on pedestrian, cyclist, and driver


behaviour of the various distances between the speed cushion and
the crossing in the different directions of the approaching vehicles.
The pedestrian crossings that were analysed were those nearest to
the speed cushion, and the distance between speed cushion and
pedestrian crossing is expressed as long or short (PCR = pedestrian
crossing). In cases where signicant differences were shown for the
respective road user group analysed, further analyses with tests to
see if the differences also applied per site were carried out to ensure Fig. 3. Diagram of crossing with cycle passage on Hultagatan. The shorter distance
that the difference can be considered to apply generally. is 3.0 m to the left in the gure, and the longer distance to the right is 8.0 m.
C. Johansson et al. / Accident Analysis and Prevention 43 (2011) 18461851 1849

Table 2
Free vehicle speeds (km/h) 12 m before and at the pedestrian crossing depending on short or long distance between speed cushion and pedestrian crossing (PCR = pedestrian
crossing).

Shorter distance Longer distance

12 m before PCR On PCR 12 m before PCR On PCR

Dragonstigen Mean 24.2 23.1 17.0 22.1


S.D. (mean) 1.51 1.21 0.59 0.51
90% 34 30 23 27
n 37 34 52 53
Tessins Vg Mean 23.6 23.9 23.8 22.5
S.D. (mean) 0.68 0.65 0.54 0.43
90% 29 30 30 28
n 39 45 117 132

approaching crossing compared to when the speed cushion is sit- 4.2. Results from video recording of road user behaviour
uated at a longer distance. However this hypothesis could not be
veried as drivers head movements or eye xations were not stud- 4.2.1. Pedestrians
ied. Interviews with school children at a nearby school revealed that In all, 1350 pedestrians were observed, but only 802 (approx-
the children experienced that the speeds were higher at the part imately 60%) of these encountered a vehicle; and 89% of all
of the crossing where the distance between the crossing and the pedestrians crossed the road at the pedestrian crossing. The only
speed cushion were shorter. Some children expressed that it was signicant difference shown was that pedestrians were more often
difcult to know if the driver would stop or not. A longer distance given way by the rst driver from the left if there was a greater dis-
should make it easier to distinguish if a driver is going to stop or tance between the crossing and the speed cushion, 50% at greater
not (Leden et al., 2006). distances compared with 40% at shorter (p < 0.05), see Table 4.
Speeds 12 m before the speed cushion were also measured, and Based on all 255 pedestrian observations from Dragonstigen, the
the results are presented in Table 2. The speed 12 m before the tendency was the same, 50% compared with 43% (difference not sig-
pedestrian crossing was comparable with that on the crossing, with nicant), on Hultagatan (187 observations) there was a signicant
the exceptions of one direction at Dragonstigen where the begin- difference of 43% and 23% (p < 0.01). On Tessins Vg, based on a total
ning of the speed cushion is 12 m before the pedestrian crossing. of 74 observations, the proportion given way by the rst driver from
The fact that speed 12 m before the crossing was the same or less left was 64% when the distance was long, and 70% when the distance
than on the crossing can most plausible be explained by drivers was short (difference not signicant), i.e. just the opposite for the
starting to decelerate in order to drive over the speed cushion, whole group of pedestrians in total, and for the other places. This
and then starting to accelerate again once they had reached the may possibly be due to lower visibility at the short distance than at
crossing. Vehicle speeds are assumed to be lowest at the speed the longer distance.
cushions. There was also a signicant difference between the three sites
On the comparison sites at Regementsgatan/Sergels Vg and in terms of the proportion given way by the rst driver from the
Regementsgatan/Skvadronsgatan (also equipped with speed cush-
ions at a shorter distance), speeds at the pedestrian crossing were
Table 4
somewhat lower than at the sites above. On neither of the study Pedestrian and driver behaviour depending on distances between left speed cushion
sites was the 90th percentile higher than 30 km/h at the cross- and pedestrian crossing. PCR = pedestrian crossing.
ing. Vehicle speeds 12 m before the crossings were comparable
Total Percentage
with the sites above. However the speed differences between 12 m number of with observed
before and at the pedestrian crossing were somewhat higher at the observations behaviour
comparison sites than at the test sites, see Table 3.
Given way by any driver
Speed measurements at Hultagatan were carried out using radar Shorter distance left speed cushion to PCR 400 71
and at the time of measurement did not differentiate between the Longer distance left speed cushion to PCR 325 74
two driving directions, which did not give sufcient accuracy for Given way by any driver from the left
stating speed at the pedestrian crossing or in different directions. Shorter distance left speed cushion to PCR 292 58
Longer distance left speed cushion to PCR 224 60
Average speed was 32.4 km/h (standard deviation for the mean Given way by rst driver from the left a

is 1.5 km/h, n = 22, 90th percentile was 40.9 km/h). Speeds were Shorter distance left speed cushion to PCR 292 40
higher at the Hulta site because the speed cushions are lower than Longer distance left speed cushion to PCR 224 50
on the sites in Malm. Run before arriving at the kerb
Shorter distance left speed cushion to PCR 627 0.5
Longer distance left speed cushion to PCR 542 1.3
Run before arriving at the refuge
Table 3 Shorter distance left speed cushion to PCR 627 1.9
Free vehicle speeds (km/h) 12 m before and at the pedestrian crossing at comparison Longer distance left speed cushion to PCR 540 3.9
sites with speed cushions at a shorter distance between speed cushion and PCR, i.e. Pedestrian stops at edge of the pavement when rst driver comes from
5 m (PCR = pedestrian crossing). the left
Shorter distance left speed cushion to PCR 181 29
12 m before PCR On PCR
Longer distance left speed cushion to PCR 193 28
Regementsgatan/Sergels Vg Mean 22.4 19.2 Pedestrians who encounter rst driver from the left look in both directions
S.D. (mean) 0.57 0.59 at the kerb
90% 28 24.0 Shorter distance left speed cushion to PCR 98 49
n 72 56 Longer distance left speed cushion to PCR 143 50
Regementsgatan/Skvadronsgatan Mean 24.3 21.6 Pedestrians who encounter any driver look in some direction at the kerb
S.D. (mean) 0.90 1.43 Shorter distance left speed cushion to PCR 98 93
90% 32 28 Longer distance left speed cushion to PCR 143 93
n 37 25 a
Signicant difference with 95% probability.
1850 C. Johansson et al. / Accident Analysis and Prevention 43 (2011) 18461851

left at short distances between speed cushion and the pedestrian Table 5
Cyclist and driver behaviour depending on distance between left speed cushion and
crossing (p < 1.0e6). There was no signicant difference between
pedestrian crossing (CCR = cycle crossing).
the sites when the distance between speed cushion and pedestrian
crossing was long. The other types of road user behaviours were Total Percentage
number of with observed
similar independent of different distances between speed cush-
observations behaviour
ion and pedestrian crossing for the other studied parameters, see
Table 4. Given way by any drivera
Shorter distance left speed cushion to CCR 167 57
No signicant differences were shown in the study of road user
Longer distance left speed cushion to CCR 173 75
behaviour in the other lane, i.e. after the refuge (in other words, Given way by any driver from the left
vehicles that came from the right seen from the pedestrians per- Shorter distance left speed cushion to CCR 128 47
spective). Driver behaviour towards the pedestrian appears to be Longer distance left speed cushion to CCR 127 57
Given way by the rst driver from the lefta
similar irrespective of whether the pedestrian is at the edge of the
Shorter distance left speed cushion to CCR 128 37
pavement or on the refuge, with about 60% of all pedestrians given Longer distance left speed cushion to CCR 127 51
way irrespective of whether the driver came from the right or left. Travel fast before arriving at the kerb
Shorter distance left speed cushion to CCR 274 0.0
4.2.2. Children and elderly pedestrians Longer distance left speed cushion to CCR 306 0.7
Travel fast before arriving at the refuge
In all, 371 children were observed, 224 (60%) of whom encoun-
Shorter distance left speed cushion to CCR 274 0.0
tered a driver, and 147 did not. Of all child pedestrians, 92% crossed Longer distance left speed cushion to CCR 306 0.7
the road at the crossing. About 1011% of all children ran over the Cyclist stops at edge of the pavement when rst driver comes from the left
second lane when crossing the street. No signicant differences Shorter distance left speed cushion to CCR 116 34
Longer distance left speed cushion to CCR 90 42
were shown in the childrens behaviour or vehicle driver behaviour
Cyclist who encounters rst driver from the left looks in both directions at
in the different directions. the kerb
A total of 258 elderly pedestrians were observed, 161 (62%) of Shorter distance left speed cushion to CCR 102 54
who encountered a vehicle; 96% of all elderly pedestrians crossed Longer distance left speed cushion to CCR 72 63
the road at the pedestrian crossing. No signicant differences were Cyclist who encounters rst driver from the left looks in some direction at
the kerb
observed in the behaviour of either the elderly pedestrians or the
Shorter distance left speed cushion to CCR 102 93
vehicle drivers from either direction. Longer distance left speed cushion to CCR 72 94
a
Signicant difference with 95% probability.
4.2.3. Bicyclists
For cyclists the analyses were carried out in the same way as for
pedestrians, but for cyclists the road user behaviour is compared crossing was longer, 80% as opposed to 51% at shorter distances
with long and short distances between speed cushions and the cycle (p < 0.001). Again, like the all cyclists group, cycling children at
crossing. However on the Tessins Vg site there is no cycle cross- the refuge were also more often given way when the distance was
ing. The bicyclists cross on the pedestrian crossing, so distances are shorter, 78% as opposed to 58% at longer distances. Only 33 elderly
measured between the speed cushion and the pedestrian cross- cyclists were observed, which is too small for further analysis.
ing. On Hultagatan the distance is approximately equal between
the cycle crossings and the two speed cushions in each direction 5. Conclusions and discussion
(5 and 6 m respectively). If the measurements from Hultagatan are
excluded the results are similar. Vehicle speeds were somewhat lower at the pedestrian cross-
A total of 613 cyclists was observed, and 356 (58%) of them ing when the distance between the speed cushion to the pedestrian
encountered a driver. A total of 84% of all cyclists crossed the road crossing was greater, and there were positive aspects regarding the
at the pedestrian crossing or cycle crossing. mobility of the pedestrians and cyclists, a greater distance of about
Cyclists were given way by drivers signicantly more often 10 m or two car lengths between the hump and the pedestrian
when the distance between speed cushion and cycle crossings was crossing is suggested.
longer, 75% compared to 57% (p < 0.01), see Table 5. Cyclists were Average motor vehicle speed was approximately 2223 km/h on
also signicantly more often given way by the rst driver (p < 0.05), the study sites, and 90th percentiles were around 30 km/h, which
51% compared to 37%. Based on a total of 231 cyclist observations means that the objective of the layouts, motor vehicle speeds below
from Dragonstigen, the tendency was the same, 82% compared to 30 km/h was met. A longer distance between speed cushion and
58% (p < 1.0e04), and on Hultagatan (96 observations) 62% and pedestrian crossing still produced lower speeds. Since the speed
57% (not signicant). On Tessins Vg, only 13 observations were cushion at one site was lower, speeds were higher, with an average
made in total. of over 30 km/h. In general, however, it is reasonable to assume that
Cyclists at the refuge were also signicantly more often given vehicle speed on a pedestrian crossing depends mainly on speed
way by some driver when the distance to the speed cushion was cushion height, provided the distance to the speed cushion is not
shorter, 74%, compared to 57% for longer distance (p < 0.01). A plau- too great.
sible explanation is that the cyclists in question were given way by All pedestrians were given way by the rst driver from the left
the rst driver they encountered, which is assumed to be from the signicantly more often; all adult and child cyclists were more often
left, and then it is reasonable that also drivers from the right give given way by drivers from the left, when the distance between the
way, since those from the left have already waited or slowed up. speed cushion and the pedestrian or cycle crossing was greater.
That behaviour was not detected in the case of pedestrians. It should be remembered that it is the cyclists who should give
way to other vehicles when they cross the road at a cycle crossing. It
4.2.4. Cycling children and elderly is also important to remember that what is a long distance to pedes-
In all, 134 cycling children were observed, and 86 (64%) of them trian crossing can be a short distance for cycle crossing when the
encountered a vehicle and 92% of all cycling children crossed the both areas often lie parallel to one another. More data to describe
road at the pedestrian or cycle crossing. cyclist safety and accessibility on cycle crossings is needed.
As for all cyclists, cycling children were more often given way Since motor vehicle speeds were low on the pedestrian crossings
when the distance between the speed cushion and pedestrian irrespective of the distance between cushion and crossing, i.e. one
C. Johansson et al. / Accident Analysis and Prevention 43 (2011) 18461851 1851

or two car lengths, and that it has been shown to have certain pos- Johansson, C., 2001. Towards a Method of Improving Road Safety for Pedestrians
itive effects concerning mobility for unprotected road users when and Cyclists, Especially in Child Pedestrian Environments. A Case Study in Bors,
Sweden, Lule Technical University. LIC, p. 29.
the distance was two car lengths, it is suggested that this distance Johansson, C., 2004. Safety and Mobility of Children Crossing Streets as Pedestrians
should be used when installing new speed cushions. A longer dis- and Bicyclists. Doctoral Thesis. Lule University of Technology, p. 27.
tance should also make it easier especially for children and elderly Johansson, C., Leden, L., 2007. Short-term effects of countermeasures for improved
safety and mobility at marked pedestrian crosswalks in Bors Sweden. Accident
to distinguish if a driver is going to stop or not, which increase Analysis & Prevention 39 (3), 500509.
security and mobility. Leden, L., Andersson, K., Kllstrm, L., 1982. Speed Reducing Devices in residential
Areas. The Swedish Roas Safety Ofce, report nr 4.
Leden, L., Johansson, C., Leden, L., 2006. Skolenkt och intervjuer om ombyggnaden
Acknowledgments av Tessins vg vid Fridhemsskolan i Malm i maj 2006 - ett underlag fr att
utvrdera effekten av ombyggnaden av Tessins vg. Lule University of Tech-
This survey was nanced by Skyltfonden (the Swedish Road nology, Arbetsrapport, p. 5.
Leden, L., Grder, P., Johansson, C., 2008. Safe Pedestrian Crossings for Children and
Administrations Road-Sign Fund). The work took place with the Elderly. Accident Analysis and Prevention 38 (6), 289294.
kind cooperation of Malm Highways Department, with special vstedal,L., Ryeng, E. (1999) Registrering av barns atferd p skolevei. SINTEF rapport
thanks to Olle Thor and Karin Caesar. It is our hope that the survey STF22A99556, Trondheim.
Siegel, S., Castellan, N.J., 1988. Nonparametric Statistics for the Behavioral Science.
will provide a useful basis for safety and accessibility for pedestri-
McGraw-Hill Book Company, New York.
ans and cyclists, especially children and the elderly. Teichgrber, W., 1983. The signicance of speed for trafc safety. Journal of Trafc
Safety 2 (II Quarter).
Vrhelyi, A., 1998. Drivers speed behaviour at a zebra crossing: a case study. Acci-
References
dent Analysis & Prevention 30 (6), 731743.
Waltz, F.H., Hoeiger, M., Fehlmann, W., 1983. Speed limit reduction from
Ashton, S.J., 1982. In: Chapman, A.J., Wade, F.M., Foot, H.C. (Eds.), Vehicle Design and 60 to 50 km/h and pedestrian injuries. In: Twenty-Seventh Step Car Crash
Pedestrian Injuries. I. Pedestrian Accidents. John Wiley & Sons Ltd. Conference Proceedings. International Research Council on Biokinetics of
Barbosa, H., Tight, M.R., May, A.D., 2000. A model of speed proles for trafc calmed Impacts (IRCOBI) , Society of Automotive Engineers, Warrendale PA, pp.
roads. Transport Research Part A 34, 103123. 311318.
Elvik, R., Vaa, T., 2004. The Handbook of Road Safety Measures. Elsevier Science Ltd. Vgverket, 2004. Vgars och gators utformning. VGU. VV Publikation, p. 80. ISSN
based on Elvik, R., Mysen, A.B., Vaa, T., 1997. Trafc Safety Handbook. Institute 1401-9612.
of Transport Economics, Oslo.