Académique Documents
Professionnel Documents
Culture Documents
ABSTRACT This paper discusses issues that are encountered in the modelling of the operational behaviour of pedestrians such as the representation of pedestrian street crossing
movements and speed choice at a micro-scale. A comprehensive literature review is undertaken for various parameters of pedestrian movement that are of fundamental importance
in any pedestrian modelling approach. These parameters are pedestrian speeds, pedestrian
speedflowdensity relationships, pedestrian compliance to traffic signals, and pedestrian
gap acceptance while crossing the road. Based on the research evidence from the literature,
a modelling framework for examining pedestrian speed choice is presented that postulates
that pedestrian speed is a function of a pedestrians value of time, risk and capabilities.
Introduction
Since the late 1980s, transport policy in many countries has sought to increase the
share of walking as a mode of transport and to increase the level of pedestrian
activity. There are many reasons for this policy goal, ranging from efforts to
curtail growth in car usage to reduce traffic and congestion and the associated
environmental impacts, to enhancing urban livelihoods, to more recent concerns
about obesity in the population and amongst children in particular. Underlying
the policy environment has been a need to understand the elements of pedestrian
behaviour and what specific policies and design measures can encourage
increased walking.
Representation of pedestrians in traffic models, in terms of the accuracy with
which their movements can be realistically simulated, is an area that has not
received much attention owing to the lack of demand from the end users of such
models, the urban transport planners. Models of traffic movements in urban areas
have traditionally ignored pedestrians. Batty (2001) gives two basic reasons for
this omission. First is the origin of transport planning as an institution in itself.
This institution owes its birth to large-scale investments in inter-urban highways
Correspondence Address: Robert B. Noland, Department of Civil & Environmental Engineering, Centre
for Transport Studies, Imperial College London, London SW7 2AZ, UK. Email: r.noland@imperial.
ac.uk
0144-1647 print/1464-5327 online/08/010061-25 2008 Taylor & Francis
DOI: 10.1080/01441640701365239
62
in the USA in the 1950s, which were in response to the rapid increase in car
ownership. The investments spread to cover urban areas, but the original focus on
motorized transport continued and in many countries continues to this day. The
second reason cited by Batty (2001) is the scale of planning. Traffic models have
traditionally focused on analysing traffic movements, modal splits, and origins
and destinations that involve large numbers of people moving to and from large
transportation zones. The scales used in such models never reach the level where
detailed investigation of walking trips is feasible.
A range of methods have been used by researchers to study pedestrian behaviour and analyse measures of effectiveness, such as delay. The choice of technique
employed for such studies has depended on the level of behaviour under
investigation. Pedestrian behaviour has been classified at three different levels
(Hoogendoorn and Bovy, 2004; Daamen, 2004) (Figure 1).
At the strategic level the pedestrian decides what activities to perform and
whether to access those activities via walking. Short-term decisions are made at
the tactical level in order to fulfil the objectives set at the strategic level. These
short-term decisions include the order in which activities are to be performed,
where these activities are to be performed, and which route to use. The expected
choice set at the tactical level also affects the decisions taken at the strategic
level and hence the decision-making process is not strictly segregated but takes
place as a two-way communication between these two levels of decisionmaking.
Much research recently has sought to include pedestrians in a mode choice
model and to consider the micro-scale impacts of urban form and design (Cervero
and Kockleman, 1997). In practice it has become increasingly common to attempt
to qualitatively assess the pedestrian environment as an attribute in mode choice
models, such as the Pedestrian Environment Factor designed for the Portland
Land Use Transportation Air Quality Connection (LUTRAQ) studies in the 1990s
Figure 1 Levels in pedestrian behaviour (Daamen, 2004)
Figure 1.
64
66
15 percentile speed
(m/s)
na
Adults
Elderly
Adults
Elderly
1.14
7123
263
1947
2445
1.51
1.35
1.44
1.45b
1.44b
1.11
1.13
1.16
1.03
1.27
1.25
0.97
1.16
1.34b
1.03
75 000
608
1.44b
1.57
1.32
1.47
1.45
1.25
1.00
1.22
1.17
1.00
0.97
0.67
0.92
0.97
Sample size.
Median speed.
1 m/s = 3.6 km/h 2.25 mph.
b
Table 1 shows that most of the studies conducted in the last few years have
reported results, not just for mean values of speed, but also the 15 percentile
values and sometimes for the entire range of speed distribution. This is due to the
fact that mean speeds are not always the best criteria for safer designs especially
for signal controlled elements of road infrastructure, and in particular if design
criteria are aimed at minimizing the risk of the slowest pedestrians.
The studies from the UK (Cresswell et al., 1978; Wilson and Grayson, 1980;
Griffiths et al., 1984) have reported a mean speed of elderly pedestrians between
1.11 and 1.16 m/s. There is more variation in younger adult pedestrian speeds,
where mean speeds range between 1.32 and 1.57 m/s. An added complication in
comparing results for adults is the different definition of adults used in the three
studies. Cresswell et al. (1978) and Griffiths et al. (1984) have defined three age
groups; young, adults, and elderly; but have not defined the criteria they have
used. The mean speed of young pedestrians reported by Griffiths et al. is 1.72 m/
s (higher than that of adults) compared with 1.47 m/s (lower than that of adults)
reported by Cresswell et al. (1978). This adds to the uncertainty of what age
group constitutes young. The statistical significance of the difference is also not
given.
Cresswell et al. (1978) collected data in Cardiff on three pelican (press-button
signal activated) crossings using video recordings. The work of Wilson and Grayson (1980) is based on 11 111 crossing observations at one site in London and two
in Southampton during off-peak hours using time-lapse photography. These
observations were carried out for random crossings and the sample size was
reduced to include only individual pedestrians who crossed the road straight and
did not experience any delay at the curb. Griffiths et al. (1984) collected data from
direct observations at 48 pelican crossings and 26 Zebra crossings in a number of
cities in England and Wales. The size of samples, type of pedestrian crossings and
land use in which their study was carried out are likely to have influenced the
results. However, due to lack of detailed results from Cresswell et al. (1978) and
68
Griffiths et al. (1984) it is not possible to draw any definite conclusions on how
these attributes affected speeds.
Wilson and Grayson (1980) have examined the relationship of pedestrian speed
with age and gender. No statistically significant difference was found in the mean
speeds of male (1.32 m/s) and female (1.27 m/s) pedestrians but when the data
was segregated into various age groups, noticeable differences were discovered.
Male teenagers had higher speeds in comparison with females; however, a
decline in speeds with age was found to be more pronounced in males than in
females. Above 60 years the mean speeds decreased to values below 1.2 m/s for
both genders, and above about 70 years of age, males were reported to be slower
than females.
Although only mean pedestrian speeds for the elderly are available from the
studies in the UK, the results are still important on three accounts. First, since
elderly pedestrians are likely to maximize their walking pace when crossing
streets (Bailey et al., 1992), their estimated speeds on crossings is likely near the
higher end of their capabilities. Second, the current UK government guidelines on
calculating the clearance interval following the pedestrian phase in a signal cycle
are based on a speed value of 1.2 m/s (Department for Transport (DfT), 2005).
This specified value is clearly insufficient from an elderly pedestrians perspective. The third important issue is that the guideline limit of 1.2 m/s has been in
use for a long time (Highways Agency, 1981) which shows a lack of consideration
given increases in the ageing population. Long cycle times that result in a large
delay to pedestrians might only cause frustration to a younger pedestrian, but an
elderly pedestrian would additionally find standing and waiting very tiring and
uncomfortable (Simes and Marin-Lamellet, 2002).
Studies from the USA (Bowman and Vecellio, 1994; Knoblauch et al., 1996; Guerrier and Jolibois, 1998; Gates et al., 2006; Fitzpatrick et al., 2006) show a greater variation in the road crossing speeds of the elderly compared with those found in the
UK studies. The mean speeds for elderly pedestrians vary between 0.97 and 1.27
m/s while the 15 percentile values are between 0.67 and 0.97 m/s. The mean adult
pedestrian speeds, as listed in Table 1, range between 1.35 and 1.51 m/s for the US
studies. This range is similar to those from the UK studies (1.321.57 m/s). The 15
percentile of adult speeds from the USA is in the range 1.001.25 m/s.
Relatively low pedestrian speeds are reported by Guerrier and Jolibois (1998)
who conducted their research on pedestrian speeds in the city of Miami Beach,
Florida. Data on pedestrian speeds was collected at five intersections using video
recording with 263 observations obtained including 40 for those observed to be 65
years or older. It was observed that clearance intervals at all intersections were
sufficient for the mean speeds of the elderly pedestrians but for the 15 percentile
values only one intersection provided adequate time. In a simultaneous questionnaire survey (not necessarily for the same sample for which speeds were
recorded) 29% of those surveyed were tourists, 31% reported having problems
with sight, and 33% reported problems with walking. This could explain the
lower mean speeds recorded in their study.
Coffin and Morrall (1995) based on their study in Canada, recommend a design
speed of 1.0 m/s to be used at mid-block crossings where there are a large
number of elderly pedestrians while Fitzpatrick et al. (2006) after a detailed field
study and a comprehensive speed comparison from the literature, recommended
a general design speed of 1.07 m/s to be reduced to 0.9 m/s where greater
consideration of elderly and pedestrians with walking problems is required.
70
affected by the delay associated with various crossing types and traffic flows. An
early analysis by Goldschmidt (1977) examined factors associated with pedestrian
delay at crossings and found that older people and children generally are delayed
longer than adults and that factors that influence delay include traffic flow, traffic
arrival patterns, the number of heavy vehicles, speed and road width.
Crompton (1979) analysed how crossing speeds vary with pedestrian delay at
randomly selected crossing points in London (Figure 2). As delay at the crossing
increases, the speed with which pedestrians cross the road also increases. This
increase is most noticeable when the pedestrian delay increases beyond 15 s.
Assuming that all the delay is caused by the lack of acceptable gaps in traffic, the
result indicates that if pedestrian delay increases beyond a certain threshold (15 s
in this case), pedestrians might perceive this delay to be excessive and try to
compensate for the lost time by walking faster than they otherwise would have.
The higher speeds when delayed could also reflect annoyance at being delayed
and a willingness to accept shorter gaps in traffic by walking faster.
Crompton (1979) also measured mean speeds for various crossing types. At
Zebra crossings (where pedestrians have priority over vehicles) and signal
controlled intersections mean speeds were 1.49 m/s. Mean speeds at crossings
with a pedestrian refuge were 1.71 m/s, at pelican crossings 1.74 m/s, and at
random crossings 1.8 m/s. Results were not segregated for age, gender and group
size and the variability of data was also not reported. In the same study mean
pedestrian delay was found to be lowest at Zebra crossings (4.9 s) followed by
random crossings (8.7 s), pedestrian refuges (12.8 s), signal controlled intersections (14.4 s), and the highest at pelican crossings (19.9 s).
Zebra crossings had both the lowest crossing speed and the lowest delay. Alternatively, pelican crossings have one of the higher crossing speeds and the greatest
delay. Both crossing types require the traffic to stop when the pedestrian crosses,
thus one would not expect the risk of crossing to affect the crossing speed.
Random crossings appear to lower mean delay, but have a higher speed, to
compensate for the lack of protection from vehicles; crossings with pedestrian
Figure 2 Mean pedestrian speed while crossing a road: relationship with pedestrian delay on a random crossing point (based on results from Crompton, 1979)
Figure 2.
Mean pedestrian speed while crossing a road: relationship with pedestrian delay on a
random crossing point (based on results from Crompton, 1979)
72
was higher at 6.2 and 4.8%. A similar effect was found by Das et al. (2005) who
found that those waiting on pedestrian refuges (or medians) are more willing to
accept shorter gaps, than those beginning the first stage of a road crossing. This
could perhaps indicate that those who are willing to wait on a median refuge
(which may be a risky place to wait) are greater risk-takers overall, or perhaps are
more time constrained and thus accept a greater level of risk.
Alternatively, DiPietro and King (1970) found that as the waiting time on the
curb increases the accepted gaps become longer. In other words the pedestrians
with longer delay at the curb are those who accept (or need) longer gaps in traffic
to cross the road. The results also show that male pedestrians accepted shorter
gaps than female pedestrians and groups of pedestrians accepted shorter gaps
than individuals. This latter result is somewhat surprising, since pedestrians in
groups have slower speeds than individuals. This finding would probably be
indicative of peer pressure to cross or lack of attention, but also might be a result
of a perception of safety in numbers. The minimum gaps accepted were 3 s for
nearside traffic and 10 s for both nearside and farside traffic. The far greater gap
needed in two way traffic streams shows the importance of pedestrian refuges in
reducing pedestrian delay.
Das et al. (2005) analysed revealed preference data based on video recordings of
a crossing in India. They estimate an ordered probit model of gap acceptance.
One of their key findings is that children and younger people accept gaps that
older people reject, but find no difference due to gender. Their results also
suggest that at signalized intersections, pedestrians may become more patient as
they wait longer, since they know the signal will change sooner.
Pedestrian crossing speed also depends on what stage of a signal cycle the
pedestrian arrives at the road crossing. Those arriving during the clearance
interval that follows the pedestrian phase may still attempt to cross the road,
by increasing their speed, rather than choosing to wait for the next pedestrian
phase (Virkler, 1998a). Lam et al. (1995) found pedestrian speeds are higher
during the pedestrian red phase in comparison with the pedestrian green
phase in Hong Kong (1.5 versus 1.27 m/s). Similarly Gates et al. (2006) found
higher speeds for pedestrians crossing outside of the pedestrian green phase
(1.52 m/s) in comparison with those crossing during the pedestrian green
phase (1.37 m/s).
Pedestrian non-compliance with crossing regulations is dependent on the
relative risk of non-compliance. Studies have found that when the width of
the carriageway is smaller or when there is a median refuge, non-compliance
rates increase. Table 2 lists the results of a variety of studies that examined
non-compliance rates. In general, these studies find significant variation
depending on both the details of the road layout and signal type, but also
some variations by gender and age of the pedestrian. The type of non-compliance also varied. Sometimes this involves crossing against a flashing signal
(temporal non-compliance) and frequently it involves crossing near but not on
the designated crossing point (spatial non-compliance). Non-compliance is
certainly a strategy that pedestrians use to reduce their overall travel time. For
example, Virkler (1998b) found that estimated pedestrian delay is reduced by
about 22%.
Chu et al. (2004) estimated a model for choice of street crossing, based on statedpreference data collected in Florida and found that crossing distance (road width)
and extra walking distance required to reach an intersection were important
Location
UK
UK
Ohio, USA
Brisbane, Australia
Texas, USA
Kansas, USA
Michigan, USA
Xian, China
Study
Rouphail (1984)
Virkler (1998b)
Stephanie and
Machemehl (1999)
Eustace (2001)
367%
917,
3744%
15%
Non-compliance rate
signalized intersections
signalized crossing
at signalized intersections
Crossing type
Comments
74
Study location
Mean speed
Older (1968)
Hoel (1968)
London
Pittsburgh,
USA
Leeds, UK
1.3 m/s
1.5 m/s (male), 1.41 m/s (female)
Singapore
OFlaherty and
Parkinson (1972)
Tanaboriboon et al.
(1986)
Tanaboriboon and
Guyano (1991)
Bangkok,
Thailand
York and
Edinburgh, UK
76
Study
Context
Mean speed
1.61 m/s
1.53 m/s
Fruin (1987)
laboratory study
It is not clear from the study if these results are for groups or individuals or both.
Study
and Elayadath (1994) with spectators moving slowly due to the non urgency of
their trip (Figure 4). Navin and Wheeler estimated a linear relationship between
speed and density; whereas Virker and Elayadath proposed a two regime
relationship (for densities lower and higher than 1.07 ped/m2). Polus et al. (1983)
also compared one- and three-regime models (for densities in the range of < 0.6,
0.60.75 and > 0.75 ped/m2). The three regime model gave a better fit with the
observed data.
Unlike Older (1968), Navin and Wheeler (1969) found loss of capacity due to bidirectional flow. At densities of 1.07 ped/m2 they estimated a loss of flow
capacity of 4% (compared with unidirectional flow) with bi-directional flow ratios
of 50:50 that increased steadily to a 14.5% loss of flow capacity with bi-directional
flow ratios of 90:10. The unidirectional capacity of footways was estimated to be
5100 ped/m/h.
Fruin (1971) proposed two linear models for uni- and bi-directional pedestrian
flows based on data collected in a bus terminal in New York City with high
commuter traffic. Separate models also suggested some loss of capacity due to bidirectional flow. The curves exhibit somewhat higher speeds at lower densities
than for indoor walkways at a train station as found by Lam et al. (1995) in Hong
Kong. It is difficult to compare these two results with those found by Hankin and
Wright (1958) for London Underground stations because of the use of different
modelling approaches (linear versus non-linear); however, at lower and higher
density the Fruin model is very close to that of Hankin and Wright. Fruin found
maximum flow volumes of 5150 and 4850 ped/m/h for uni- and bi-directional
78
Figure 3. Pedestrian speeddensity relationships from various studies in the UK. All figures are
based on plotting the formulas in Table 5, except for Hankin and Wright (1958), which was redrawn
based on their figures 1 and 3; and Brocklehurst et al. (2005) based on their figure 8
flows, respectively, values slightly less than the design limit of 5300 ped/m/h
suggested by Hankin and Wright. Lam et al. found a much lower capacity for
indoor walkways at 4080 ped/m/h.
For shopping streets the model proposed by Tanaboriboon et al. (1986) based on
data from Singapore and by Polus et al. (1983) from Israel are similar. They also
match Olders model at lower densities. Lam et al. (1995) also proposed two nonlinear models for pedestrian signal control crossings both of which diverge from
each other at lower and higher densities but give the same results for an intermediate range of densities. The capacity of the two crossings is 3600 and 5400 ped/
m/h, which is a significant difference. The reasons for the difference in the
models and capacities are not clear.
Al-Masaeid et al. (1993) collected data on pedestrian walking characteristics
in Jordan and found that some pedestrians walked on carriageways even at
lower densities; however, the proportion walking on the carriageway increased
substantially as the densities on the sidewalk increased. It was also found that
bi-directional flow caused capacity (measured to be 3590 ped/m/h) to
decrease as much as 25% from the maximum unidirectional capacity. The
model proposed by Sarkar and Janardhan (2001) bears close resemblance to
that of Al-Masaeid et al. (1993) even though one is linear and the other nonlinear. This suggests that relationships in developing countries could have
different effects than in developed countries. Clearly the relative levels of
walking versus car usage, as well as the suitability of pedestrian infrastructure
plays a role.
Figure 4.
Pedestrian speedflow relationships from studies outside of the UK. All figures are based
on plotting the formulas in Table 5
80
82
behaviour and the role of numerous variables that influence this behaviour. At
the microscopic level, individual pedestrian movements are dependent on both
individual characteristics, the situation and type of trip being made, the presence
of other pedestrians, whether the pedestrian is in a group, and the road crossing
infrastructure and characteristics of the traffic flow and speed.
This paper provides a review of some of the key empirical results on how
pedestrians choose their speeds and cross the road. These results are interpreted
to formulate a theoretical model that can explain the various trade-offs made by
individuals. This is formulated as a choice of speed based on individual capabilities, value of time, and the risk of the specific situation.
Consideration of these trade-offs can provide a useful framework for analysing
pedestrian policy proposals and for understanding the key trade-offs that individual pedestrians make. It is hoped that an understanding of these issues will
increase the awareness and sensitivity of transport planners and engineers to
these issues in specific site designs, rather than focusing on prescriptive
standards. This framework may also provide a useful guide for further improving
pedestrian modelling techniques.
Acknowledgements
The authors acknowledge the support from the Rees Jeffreys Road Fund, for
which they are grateful.
References
1000 Friends of Oregon (1993) The Pedestrian Environment. LUTRAQ Report Vol. 4A, prepared by
Parsons Brinckerhoff Quade & Douglas.
1000 Friends of Oregon (1996) Analysis of Alternatives. LUTRAQ Report Vol. 5, prepared by Cambridge
Systematics, Inc. and Parsons Brinckerhoff Quade & Douglas.
1000 Friends of Oregon (1997) Making the Connection. LUTRAQ Report Vol. 8, prepared by Parsons
Brinckerhoff Quade & Douglas.
Al-Masaeid, H. R., Al-Suleiman, T. I. and Nelson, D. C. (1993) Pedestrian speedflow relationship for
central business district areas in developing countries, Transportation Research Record: Journal of the
Transportation Research Board, 1396, pp. 6974.
Bailey, S. S., Jones, S. A., Stout, R. J., Bailey, J. H., Kass, S. and Morgan, B. B. (1992) Issues of
elderly pedestrians, Transportation Research Record: Journal of the Transportation Research Board,
1375, pp. 6873.
Batty, M. (2001) Editorial: Agent-based pedestrian modelling, Environment and Planning, 28B,
pp. 321326.
Bennett, S., Felton, A. and Akelik, R. (2001) Pedestrian movement characteristics at signalised intersections. Paper presented at the 23rd Conference of Australian Institutes of Transport Research,
Melbourne, Australia.
Birchall, M., Barton, J. and Leather, J. (1994) PAXPORTthe airport terminal simulation model.
Paper presented at the 22nd European Transport Forum. Proceedings of Seminar B, PTRC, P374,
pp. 6175.
Boarnet, M. and Crane, R. (2001) Travel by Design: The Influence of Urban Form on Travel (Oxford: Oxford
University Press).
Bowman, B. L. and Vecellio, R. L. (1994) Pedestrian walking speeds and conflicts at urban
median locations, Transportation Research Record: Journal of the Transportation Research Board,
1438, pp. 6773.
Brocklehurst, D., Bouchlaghem, D., Pitfield, D., Palmer, G. and Still, K. (2005) Crowd circulation and
stadium design: low flow rate systems, Proceedings of the Institution of Civil Engineers: Structures and
Buildings, 158, pp. 281289.
Buchanan, C. (1963) Traffic in Towns: A Study of the Long Term Problems of Traffic in Urban Areas.
Ministry of Transport; Working Group (London: HMSO).
84
Helbing, D. and Molnar, P. (1995) Social force model for pedestrian dynamics, Physical Review, 51E,
pp. 42824286.
Henderson, L. F. (1974) On the fluid mechanic of human crowd motions, Transportation Research, 8,
pp. 509515.
Highways Agency (1981) Pedestrian Facilities at Traffic Signal Installations. Technical Advice Notes No.
TA 15/81 (London: Department of Transport).
Hine, J. and Russell, J. (1996) The impact of traffic on pedestrian behaviour 2: Assessing the traffic
barrier on radial routes, Traffic Engineering and Control, 37(2), pp. 8185.
Hoel, L. A. (1968) Pedestrian travel rates in central business districts, Traffic Engineering, 38(4),
pp. 1013.
Hoogendoorn, S. P. and Bovy, P. H. L. (2004) Pedestrian route-choice and activity scheduling theory
and models, Transportation Research, 38B, pp. 169190.
Hughes, R. L. (2003) The flow of human crowds, Annual Review of Fluid Mechanics, 35, pp. 169182.
Ishaque, M. M. (2006) Policies for pedestrian access: multi-modal trade-off analysis using microsimulation techniques. PhD dissertation, University of London, London.
Jacobs G. D. and Wilson, D. G. (1967) A Study of Pedestrian Risk in Crossing Busy Roads in Four Towns.
Report No. LR106 (Berkshire, UK: Road Research Laboratory).
Jacobs, G. D., Older, S. J. and Wilson, D. G. (1968) A Comparison of X-Way and Other Pedestrian
Crossings. Report No. LR145 (Berkshire, UK: Road Research Laboratory).
Jones, S. L., Sullivan, A. J., Cheekoti, N., Anderson, M. D. and Malave, D. (2004) Traffic Simulation
Software Comparison Study. Report No. 02217 (Birmingham, AL: University Transportation Centre
for Alabama) (available at: http://utca.eng.ua.edu/projects).
Kerridge, J., Hine, J. and Wigan, M. (2001) Agent-based modelling of pedestrian movements: the
questions that need to be asked and answered, Environment and Planning, 28B, pp. 327341.
Knoblauch, R. L., Pietrucha, M. T. and Nitzburg, M. (1996) Field studies of pedestrian walking speed
and start-up time, Transportation Research Record: Journal of the Transportation Research Board, 1538,
pp. 2738.
Krajzewicz, C. and Rssel, D. (2006) SUMO User Documentation Rev. 1.24 (available at: http://
sumo.sourceforge.net).
Kuligowski, E. (2004) A comprehensive review of 28 evacuation models. Paper presented at the
Workshop on Building Occupant Movement during Fire Emergencies (Gaitherburg, MD: National
Institute of Standards and Technology) (available at: http://fire.nist.gov/CDPUBS/NISTSP_1032/
Data.htm).
Lam, W. H. K., Morrall, J. F. and Ho, H. (1995) Pedestrian flow characteristics in Hong Kong,
Transportation Research Record: Journal of the Transportation Research Board, 1487, pp. 5662.
Liu, R. (2005) DRACULA 2.3 User Manual (Leeds: University of Leeds) (available at: http://
www.its.leeds.ac.uk/software/dracula).
Maw, J. R. and Dix, M. (1990) Appraisals of station congestion relief schemes on London
Underground. Paper presented at the 18th Summer Annual Meeting. Proceedings of Seminar H,
PTRC, P335, pp. 167178.
Moore, R. L. (1953) Pedestrian choice and judgment, OR: Journal of the Operational Research Society, 4,
pp. 310.
Navin, F. P. D. and Wheeler, R. J. (1969) Pedestrian flow characteristics, Traffic Engineering, 39(4),
pp. 3036.
Noland, R. B. (2006) Behavioural and analytical considerations in transport safety policy. Paper
presented at the 11th International Conference on Travel Behaviour Research, Kyoto, Japan.
OFlaherty, C. A. and Parkinson, M. H. (1972) Movement in a city centre footway, Traffic Engineering
and Control, 13, pp. 434438.
Older, S. J. (1968) Movement of pedestrians on footways in shopping streets, Traffic Engineering and
Control, 10, pp. 160163.
Olstam, J. J. and Tapani, A. (2004) Comparison of Car-Following Models. Report No. VTI meddelande
960A (Linkping: Swedish National Road and Transport Research Institute) (available at: http://
www.vti.se/EPiBrowser/Publikationer/English/M960A.pdf).
Peltzman, S. (1975) The effects of automobile safety regulation, Journal of Political Economy, 83,
pp. 677725.
Polus, A., Schofer, J. L. and Ushpiz, A. (1983) Pedestrian flow and level of service, Journal of Transportation Engineering, 109, pp. 4656.
PTV (2004) User Manual: VISSIM 4.0 (Karlsruhe: PTV Planung Transport Verkehr AG).
Rouphail, N. M. (1984) Midblock crosswalks: a user compliance and preference study, Transportation
Research Record: Journal of the Transportation Research Board, 959, pp. 4147.