Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

Journal of Urban Design, Vol. 13. No. 1, 81–98, February 2008

How Far, by Which Route and Why? A Spatial Analysis

of Pedestrian Preference

ASHA WEINSTEIN AGRAWAL1, MARC SCHLOSSBERG2 &

KATJA IRVIN3
1
Department of Urban and Regional Planning, San José State University, CA, USA; 2Planning,
Public Policy and Management, University of Oregon, USA; 3PMC, Rancho Cordova, CA, USA

ABSTRACT This paper reports on a survey of pedestrian trips to transit that examined
the trip lengths and route choices made by people walking to five rail transit stations in
California and Oregon. In highly motorized countries such as the US, policy-makers are
beginning to recognize that shifting some travel from auto trips to walking trips can help
the country achieve important policy objectives such as combating obesity and reducing
the air pollution and oil dependency that result from auto use. However, researchers know
very little about pedestrian behaviour and the role of the built and aesthetic environment in
influencing pedestrian trips to transit. As communities wrestle with the interconnected
issues of obesity, sprawl, and quality of life, planners need to understand how far
Americans will walk to transit and the environmental factors that influence them. This
survey of 328 pedestrians walking to rail stations, primarily on weekday mornings, found
that they were willing to walk an average of half a mile to the rail station and that
minimizing the distance walked was the most important factor influencing their choice of
route. The people surveyed also frequently mentioned safety factors as important in route
choice. Aesthetic elements of the built environment, on the other hand, were rarely
mentioned as important route choice factors. The paper concludes by using these survey
findings to recommend strategies that planners, designers, and policy-makers can use to
design successful transit and pedestrian-oriented developments.

Introduction
There is an increasing public interest in community walkability, as reflected in the
growing number of state and federal initiatives on Safe Routes to School, the new
concern over a national obesity epidemic (especially in children), and a wide
range of policy initiatives designed to convince travellers to switch from auto trips
to more environmentally sustainable bicycle and walking trips. In each of these
cases, policy makers recognize walking as a key mode of travel and believe that
increasing the number of walk trips is an important goal.
Despite the seeming simplicity of the goal, very little is known about how far
people actually walk or about how street design affects people’s willingness or
Correspondence Address: Asha Weinstein Agrawal, Department of Urban and Regional
Planning, San José State University, One Washington Square, San Jose, CA 95192-0185, USA.
Email: [email protected]

1357-4809 Print/1469-9664 Online/08/010081-18 q 2008 Taylor & Francis

DOI: 10.1080/13574800701804074
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

82 A. W. Agrawal et al.

capacity to access desired destinations on foot. Developing new tools to identify

walk trip distances is a key research challenge, as is learning to characterize the
local environment from a pedestrian point of view. This paper reports on a survey
designed to answer two primary research questions related to the topic of
pedestrian behaviour:
(1) How far do pedestrians walk to rail stations?
(2) What environmental factors do they say influence their route choice?

The next section of the paper sets the context for the research by examining related
literature. This is followed by a description of the survey methodology, then by an
analysis of the survey findings, and finally the paper’s conclusions.

Research Context
There is not a well-established literature providing firm answers to either of the
paper’s two main research questions about how far pedestrians walk and the
factors that influence their route choice. While there are many rules of thumb and
educated guesses, relatively little research exists on walking behaviour in general,
and these topics in particular. Until the mid-1990s, pedestrian behaviour was
virtually ignored in the transportation and planning literatures. In the last decade,
and especially the last five years, the topic has suddenly become popular and
many papers related to pedestrians have been published or are in progress. Much
of the new literature has come from the public health community, complementing
work done by planning and transportation researchers.
The earliest and largest body of research, which comes from the
transportation planning community, assesses the factors that influence people to
choose one mode of travel instead of another. In general, the authors of these
studies want to understand what factors would shift Americans away from solo
driving trips and towards transit, biking or walking. Because the research was
usually designed solely to discover why people choose to walk instead of drive,
most studies did not examine the distances or routes walkers travelled.
The majority of these studies claim to look at what is often neighbourhoods
‘three Ds’: density, diversity and design, although in reality the studies tend to
focus on the first of the two Ds, density and diversity of land uses. Many
researchers have concluded that residents are more likely to walk in dense
neighbourhoods that include a diverse mix of non-residential uses within a short
distance (Cervero & Gorham, 1995; Cervero & Kockelman, 1997; Greenwald &
Boarnet, 2001), although a subset of the research community remains unconvinced
that the association is very strong, except for comparisons between extremely high
and extremely low densities (Badoe & Miller, 2000; Crane, 2000).
Despite the rhetoric about the three Ds, micro-scale urban design and
environmental factors were often ignored, probably because no pre-existing
datasets captured design factors such as the presence of greenery, attractive
buildings, smooth and wide sidewalks, traffic control devices to aid pedestrians
crossing the street or the presence of heavy traffic. In North America and Europe,
from the 1970s to the 1990s there were only scattered studies on the link between
the local environment and walking, with many of these focusing on how heavy
traffic volumes discourage walkers (e.g. Carp, 1971; Lövemark, 1972; Nielson &
Fowler, 1972; City Engineering Department, Vancouver, BC, 1978; Mitchell
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

How Far, by Which Route and Why? 83

& Stokes, 1982; May et al., April 1985; Seneviratne & Fraser, 1987; Hopkinson et al.,
December 1987; Hoxie et al., 1994; Handy, 1996; Shriver, 1997). Since 2000 there has
been a burst of enthusiasm for taking on the design question more rigorously, with
a number of studies on the topic appearing in the last decade (e.g. Pikora et al.,
2002; Kirtland et al., 2003; Rodrı́guez & Joo, 2004; Ewing et al., 2005). However,
researchers have quickly discovered that pedestrian behaviour is highly complex
and difficult to study, and the existing body of research points to few consistent
findings. A recent review of the evidence linking physical activity with the built
environment concluded that there is limited evidence showing a connection
between neighbourhood design and walking, but that further research is needed
to determine whether there is truly no link or if existing research has not been
designed properly to reveal real relationships (Committee on Physical Activity,
Transportation Research Board, 2005).
There is even less literature looking specifically at how far pedestrians walk
than there is about what factors influence people to walk. The main sources of
information are the US Census, National Household Travel Survey and regional
household travel surveys. These surveys often report the number of walk trips
made, but do not necessarily include trip distances, and even when they do, the
data is often suspect (Agrawal & Schimek, forthcoming). For example, in the 2001
NHTS, surveyors recoded many walk trip distances to the nearest mile. Given that
most walk trips are quite short, this recording method makes the data almost
useless for understanding walk trip distances with any precision.
In terms of how far pedestrians walk to access rail specifically, most of the
existing data are collected by transit agencies conducting internal surveys of their
passengers. Researchers usually do not have easy access to the data because no
centralized databanks collect such surveys. In addition, these surveys usually ask
respondents to estimate the distance they walked, and thus the data may not be
highly accurate. However, one study from the mid-1990s gathered a few such
surveys from the United States and Canada and conducted a survey of light rail
riders in Calgary, Canada. The authors found that the median walking distance in
Calgary was about a fifth of a mile, although at suburban stations it was twice that
distance (O’Sullivan & Morrall, 1996).

Methodology
In the survey conducted for this research, respondents answered a series of
questions about how far and how long they walked to the station, what factors
influenced their choice of route, their attitudes towards walking, and some basic
demographic questions. They also drew on a map the route they walked and any
locations they avoided on their walk.
Surveys were distributed at five rail stations. Two were in the San Francisco
Bay Area, in California: one in San Jose (Japantown) and one in El Cerrito
(El Cerrito Plaza). The other three were in Portland, Oregon (Hollywood, Gresham
and Rockwood). The primary criteria for selecting the station areas was to find
neighbourhoods where pedestrians would have a reasonably high number of
different route options between their home and the station. It was assumed that
people would not be willing to walk more than a short distance out of their way to
find a nicer route, and so the study selected only neighbourhoods with streets laid
out in a grid pattern. With a grid street pattern, respondents could choose among
multiple routes that were all approximately the same distance. Neighbourhoods
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

84 A. W. Agrawal et al.

were also chosen where walkers would have a mix of local and collector or arterial
streets, as well as both residential and mixed-use or commercial streets.

The Stations
Japantown station, San Jose, California. The Japantown station, in San Jose,
California, is part of the Santa Clara Valley Transportation Authority’s light rail
system. The station is located in historic Japantown, an area of traditional
neighbourhoods just outside of downtown San Jose. Built environment
conditions in the area vary substantially from block to block. Several medium
and high density residential projects have been completed since 2000 to the east.
To the west, much of the land is designated for open space to protect an airport
flight path.

El Cerrito Plaza station. The El Cerrito Plaza station is part of the Bay Area Rapid
Transit system (BART), which serves four counties in the San Francisco Bay Area,
California, region. The neighbourhood around the station is laid out in a grid
street network. The area is primarily residential, with several commercial streets,
plus a large shopping centre south of the BART station. Underneath the BART
tracks runs a popular bicycle and pedestrian path. The catchment area for
potential walkers to the BART station is quite large. There are no major barriers
created by freeways or other features of the built environment, except that walkers
do face some hills in neighbourhoods rising up a moderately steep hill a few
blocks to the east.

Figure 1. Looking east to Japantown station.

Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

How Far, by Which Route and Why? 85

Figures 2 and 3. Looking east to El Cerrito Plaza station. Hollywood station, located between heavy rail
and the freeway.

Hollywood station. Hollywood and the following two stations are part of
Portland’s TriMet Max Light Rail system. They all lie east of downtown Portland.
Hollywood station sits between a freeway and a heavy rail line and pedestrians
access it from either side by a pedestrian footbridge. One side of the station
consists mainly of residential housing, with mostly residential streets closest to the
station. The other side of the station is bordered by a bus drop-off zone,
commercial and office space and a combination of multi-family and single family
residential sections. Two fairly heavily used arterials bisect the space to this side of
the station.

Gresham station. Gresham station is adjacent to a centralized bus hub, and the
two transit facilities combined are considered one of TriMet’s transit centres. The
Gresham area was developed prior to World War II, and to the south and east of
the station there is a street grid pattern typical of that era. There are no arterials or
other major roads between this residential area and the station. Outside this
gridded area there are a number of major roads, some within a quarter-mile of the
station. There are also large commercial areas and offices nearby and a mixture of
both single family and multi-family residential areas.

Rockwood station. Rockwood Station sits on a busy commercial corridor with

multi-family and single family residences adjacent to it in all directions. There are
signalized crossings at adjacent arterials, but the distances to cross are quite long.
Directly across from the westbound platform sits a large commercial lot that is
currently unoccupied, although it has become an informal park-and-ride lot.

The Surveys
Surveys were distributed at transit stations to people who walked to the transit
stop. Between one and three surveyors distributed surveys, depending on the day
and station, and they worked between 6 am and 10 am mostly on weekday
mornings, from February to May 2006. The surveyors followed a script for
consistency. The surveyors asked each person if s/he had walked to the station,
and those who responded ‘yes’ were asked follow-up questions to determine their
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

86 A. W. Agrawal et al.

Figure 4. Looking at both sides of the Gresham station.

Figure 5. The west-bound train at the Rockwood station.

eligibility for the study: (1) if they were over 18 years of age, and (2) if they would
be willing to participate in the study. Willing survey respondents received a
written survey.
At four of the stations, which had low passenger volumes, surveyors
approached all people waiting at the station and asked how they arrived at the
station. At the El Cerrito BART station, which has higher ridership, the surveyor
approached a randomly selected sample of the riders waiting on the platform.
The six-page paper survey was relatively quick to complete because there
were only a few questions per page. In addition to a set of both closed and open-
ended questions, the questionnaire included a full-page map and asked
respondents to trace the route they walked that day to the station. Surveyors
gave out pens with the surveys, and several respondents finished them on the
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

How Far, by Which Route and Why? 87

Table 1. Survey response rates, by station

Station No. completed, usable surveys Response ratea (%)

El Cerrito Plaza 120 71

Japantown 90 49
Hollywood 78 45
Gresham 15 15
Rockwood 25 23
Total 328 45

Note: aResponse rate is defined as the number of surveys returned as a proportion of the number of
surveys distributed. Some riders contacted were not given a survey because they had not walked
or refused to participate.

platform. The surveys included pre-stamped envelopes, so many people comp-

leted the survey later and dropped it in the mail.
The survey included three sections:
(1) Questions on walking behaviour, preferences, and route choice.
(2) A map inserted in the survey on which respondents were asked to trace their
walking route. Respondents were also asked to mark intersections and streets
they avoided on their walk if they had not indicated them in writing in the first
part of the survey.
(3) Basic demographic questions.
A total of 728 surveys were distributed and 328 surveys were returned by
respondents. Table 1 shows the number returned per station, as well as the
response rate per station. Almost two-thirds of the surveys (64%) came from the
two Bay Area stations; over one-third of the surveys came from El Cerrito Plaza
station and just over another quarter came from the Japantown. Of the remaining
surveys, almost a quarter came from Portland’s Hollywood station (24%), and the
Gresham and Rockwood stations in Portland provided the remaining few
surveys.
The response rate for the survey was quite high. For the total population, the
response rate was 45%. El Cerrito Plaza had the highest response rate, 71%, while
response rates from the other stations ranged from 15% to 49%. The response rate
was calculated as the number of surveys returned as a proportion of the number of
surveys distributed. Some transit riders approached by the surveyors were not
given a survey to complete because they did not wish to participate or because
they had not walked to the station.
Although some surveys had missing responses for a few individual
questions, all of the surveys were complete enough to be included in the final
dataset. However, the number of responses contributing to the results varied
slightly for each question.

Analysis of Survey Findings

This section of the paper first describes the basic socio-demographic
characteristics of the survey respondents, and then discusses the results of the
survey. Sections discuss respondents’ trip purposes, how many people stopped
along their walk and what for, how far respondents said they walked, as well as
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

88 A. W. Agrawal et al.

the authors’ own calculations of the distances they travelled, the factors that
influenced their route choices and their attitudes towards walking.

Who Were the Survey Respondents?

Table 2 summarizes some socio-demographic statistics about the survey
respondents. They were approximately half male and half female, about three-
quarters self-identified as white, and three-quarters were adults between the ages
of 30 and 59. The median household income was $60 000, and slightly over half the
respondents said they rented their home. Almost one-third of the group rarely or
never had access to a car, indicating that a fairly high proportion of the
respondents were transit-dependent.
The groups of respondents from each station were all approximately similar
to the total population of respondents, with just a few notable differences. The Bay
Area respondents were a racially diverse group, while the Portland respondents
were nearly all white. In addition, the small sample of respondents from the
Portland stations of Gresham and Rockwood had considerably lower household
incomes and, correspondingly, were more likely to rent than own their homes. The

Table 2. Demographics of survey respondents

Bay Area Portland

All Holly- Rock-

stations El Cerrito Japantown Gresham wood wood

Gender
Male 53% 49% 66% 40% 47% 52%
Female 47% 51% 34% 60% 53% 48%
Race
White 74% 68% 59% 93% 96% 86%
Black 2% 0% 2% 0% 1% 9%
Asian/Pacific Islander 15% 23% 20% 0% 1% 5%
Other 5% 4% 11% 0% 0% 0%
Mixed race 5% 5% 9% 7% 1% 0%
Age
18 –29 19% 15% 25% 23% 15% 29%
30 –39 30% 34% 26% 15% 32% 29%
40 –49 23% 20% 26% 39% 25% 13%
50 –59 20% 25% 12% 8% 23% 25%
60 þ 8% 7% 10% 15% 6% 4%
Household income
Median $60 000 $80 000 $60 000 $35 000 $70 000 $20 000
Own/rent home
Own 44% 45% 38% 29% 60% 21%
Rent 56% 55% 62% 71% 40% 79%
Driver’s licence?
Yes 84% 91% 76% 80% 86% 75%
No 16% 9% 24% 20% 14% 25%
Access to a car
Never/ occasionally 30% 16% 36% 33% 32% 67%
Most of the 70% 84% 64% 67% 69% 33%
time/always
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

How Far, by Which Route and Why? 89

Rockwood population was also highly transit dependent, with 67% saying that
they never or only occasionally had access to a car.

Trips’ Purposes and Origins

Most respondents made home-based trips to work, as would be expected given
that the sampling took place during weekday mornings. Among the full
population, 81% made commute trips, another 5% made trips to school and 8%
made personal shopping trips. This pattern was almost consistent across all the
stations, except that Japantown had fewer commute trips and considerably more
shopping trips (21%), whereas Gresham riders made fewer commute trips and
more trips to school (33%).

Stops During the Trips

The survey asked respondents if they had stopped along the way to the rail
station. If they had, follow-up questions probed the reason for the stop and its
duration. The vast majority, 87%, did not stop. Of the 13% of respondents who did
stop, about half stopped to buy food or a drink. The others stopped to buy a
newspaper, to talk to somebody, or for ‘other’ reasons. The median time for these
stops was just 3 minutes, consistent with stops made by people popping into a
retail business to make a quick purchase. The average stop time was longer, at
6 minutes, reflecting the fact that some people did stop for much longer time
periods (up to 45 minutes for the longest stop).

Trip Distances
Self-reported distances. Respondents were asked how far they had walked, in both
miles and blocks. Almost all respondents entered the number of blocks (91%), but
only 64% entered the distance in miles. (The question was formatted on the page
with a space to enter the distance in miles followed by a space to enter the number
blocks on the same line. This particular layout on the page may have encouraged
some people to think they only needed to enter the information in one form.)
For the full group of respondents, the mean reported distance was 0.58 miles
(see Table 3). Looking at how the data broke out in quartiles shows that a quarter
of people reported walking just a quarter of a mile or less, the second quartile of
people reported walking between a quarter mile and a half mile, the third quartile
reported walking between half a mile and almost a full mile (0.95 miles), and the
final quarter said they walked more than 0.95 miles. The responses clustered
around 0.25, 0.5 and 1.0 miles, indicating the tendency of people to round off
distances.

Actual distances. Respondents were asked to trace on a map the route they
walked. For the El Cerrito, Japantown and Hollywood stations, these routes were
entered into a GIS database and the information used to calculate the exact length
of each trip. The mean trip distance was just over a half mile (see Table 4), with the
shortest trip being 0.02 miles and the longest 1.88 miles. Looking at the distance
data broken into quartiles shows that a quarter of respondents walked a quarter
mile or less, the next quartile walked between a quarter and half mile, the third
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

90 A. W. Agrawal et al.

Table 3. Self-reported distance walked in miles, blocks and minutes

Bay Area Portland

All stations El Cerrito Japantown Gresham Hollywood Rockwood

Distance in miles
Mean 0.58 0.65 0.45 0.43 0.62 0.49
25th percentile 0.25 0.25 0.13 0.11 0.39 0.25
50th percentile 0.50 0.50 0.28 0.30 0.50 0.50
75th percentile 0.95 1.00 0.69 0.80 1.00 0.75
Distance in blocks
Mean 6 6 4 4 8 5
25th percentile 3 3 2 2 4 2
50th percentile 5 5 4 2 6 3
75th percentile 8 8 6 4 10 6
Time (minutes)
Mean 10 11 8 7 11 10
25th percentile 5 6 5 3 5 5
50th percentile 10 10 6 6 10 10
75th percentile 12 15 10 10 13 13

quartile walked between a half and two-thirds of a mile, and the final quarter
walked over two-thirds of a mile.

The accuracy of self-reported distances. The study was interested in learning how
accurately respondents estimated the distances they had walked. Many travel
surveys ask respondents to estimate the distances they walk, but little is known
about how accurate these estimates are. Almost half of the respondents (43%)
guessed quite accurately, within a tenth of a mile (see Table 5). However, other
guesses were highly inaccurate, ranging from up to 1.07 miles over to 0.88 miles
under the correct distance. The average guess was off by about 0.2 miles.
Percentage-wise, guesses were on average off by 45% of the distance, with 25% of
respondents guessing within 11% and half guessing within 30% of the correct
distance in miles. (Since the distances walked were short, the actual error in miles
was trivial for most respondents.) On the other hand, 25% of respondents’ guesses
were off by more than 50%, a surprisingly large error, and 10% were off by more
than 90%. Overall, almost as many people over-estimated as under-estimated,
with no conclusive pattern showing up.

Table 4. Actual distances walked

Distance (miles)

Mean 0.52
Minimum 0.02
Maximum 1.88
25th percentile 0.27
50th percentile 0.47
75th percentile 0.68
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

How Far, by Which Route and Why? 91

Table 5. Accuracy of self-reported trip distances

Accuracy of distance estimate % Cumulative %

Within 0.1 mile 43 43

Off by 0.1 to 0.25 mile 31 72
Off by 0.25 to 0.5 mile 20 94
Off by .0.5 mile 6 100

Factors Influencing Route Choices

After the survey respondents traced the route they walked on a map, the survey
asked them to identify the factors that led them to choose the route they took. The
survey addressed this issue in three steps. First, respondents were asked the open-
ended question, “What are the main reasons why you chose your route today?”
and given space to write three answers. On the next page, respondents were asked
to rank the importance of 11 potential factors that might have influenced their
route choice. The instructions read: “Below is a list of factors that other researchers
have found to influence the routes people walk along. For each one, please mark
how important it is to you”. Finally, the last open-ended question asked: “Are
there any other factors, positive or negative, that influenced your choice of route
today?” Relatively few people answered this final question, so only the results of
the first two questions are discussed below.
The first question, the open-ended one, showed that by far the most
important factor was choosing the shortest or fastest route. As shown in Table 6,
52% of respondents mentioned this as the first item in their list, and almost two-
thirds mentioned this somewhere among their three responses. An additional 9%
of respondents mentioned “convenience” as an important factor, and it may well
be that convenience was their way of expressing the same concept—choosing the
quickest route.
The second most common set of responses had to do with safety, mentioned
by 28% of respondents. Most of these related in some way to safety from traffic—
people mentioned issues such as low traffic volumes or an intersection where it
was easy to cross a large street—with only a few mentioning crime safety issues.
Although safety was a popular response, only 8% of people mentioned it as their
first item on the list; safety was somewhat more common as the second item,

Table 6. Percentage of people volunteering a factor as influencing their route choice

Question: What are the main reasons why you chose your route today?

Factor type Anywhere in list First Second Third

Shortest/fastest 64 52 10 3
Safety 28 8 14 6
Convenience 9 6 2 1
Attractive 8 2 3 2
Habit 6 3 1 2
Stopped at a business 3 2 2 0
Other 27 13 9 5
Unintelligible responses 16 9 5 3
Left blank N/A 3 50 77
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

92 A. W. Agrawal et al.

appearing here 14% of the time. Finally, very small numbers of respondents
mentioned choosing their routes based either on the attractiveness of the route
(e.g. nice landscaping or attractive buildings) or because they wanted to stop at a
particular business.
These priorities were partially validated in the next question, which asked
respondents to rate the importance of 11 different factors. As shown in Table 7,
99% of respondents rated choosing the shortest route as either very important or
somewhat important, with the bulk of those saying it was very important (82% of
respondents). This finding confirms the results of the open-ended question, where
responses related to distance predominated. Moreover, the study independently
calculated respondents’ shortest route using GIS, and respondents consistently
did choose an optimized path from home to the transit stop.
Safety considerations showed up as considerably more important in the
second question than they did in the previous, open-ended question. About half of
respondents rated as very important having traffic devices present and having
traffic drive at safe speeds; those numbers jumped considerably to 85% and 87%
when the responses of those who responded that these factors were either very or
somewhat important were combined. Women were more likely than men to rate
safety factors as important: 69% rated as very important the presence of traffic
compared to 43% of men, and 36% of women rated as very important the presence
of other people on the street compared to 12% of men.
Other factors that were rated as very or somewhat important by at least 50% of
respondents were having sidewalks in good condition; the presence of attractive
buildings, trees and landscaping; having no traffic lights where it took a long time
to cross; the presence of other people out walking; and having shops or businesses
to stop in. However, only the first two of these were rated as very important by at
least one-third of respondents. Finally, three factors rated as important by relatively

Table 7. How respondents rated the importance of factors that might influence
their route choice

Question: Below is a list of factors that other researchers have found to influence the routes people walk
along. For each one, please mark how important it is to you.

Factor Very important Somewhat important Not important

Shortest route 82% 17% 1%

Traffic devices are present 55% 30% 15%
Traffic drives at safe speeds 46% 41% 13%
Sidewalks in good condition 43% 44% 13%
Presence of attractive buildings, trees, 35% 44% 21%
and landscaping
No traffic lights where it 29% 39% 32%
takes a long time to
cross
Other people out walking 23% 37% 40%
Shops/businesses to stop in 14% 32% 54%
Shops/businesses with window to 11% 25% 65%
look in
Benches/places to sit 11% 15% 75%
Friend/neighbour along the route 7% 18% 75%

Note: Factors were ordered differently in the survey itself.

Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

How Far, by Which Route and Why? 93

few people were having shops or businesses with windows to look at; having
benches or other places to sit; and having a friend or neighbour along the route.

Attitudes towards Walking

Towards the end of the survey, respondents were asked how strongly they agreed
with a series of statements describing different reasons that they might choose to
walk. Overall, respondents had very positive attitudes towards walking, perhaps
explaining their high level of willingness to complete and return the survey. The
first two questions asked people if they liked walking and if they found walking
relaxing, and in both cases 97% either strongly agreed or agreed with the
statement (see Table 8). Another question asked respondents if they walked in
order to get exercise or health benefits, and again virtually all agreed or strongly
agreed (94%). Slightly lower percentages of people agreed that they sometimes
walk because it is the most convenient mode of travel (89%) or because it is the
cheapest way to travel (80%).

Study Findings and their Implications for Design and Planning

This study surveyed pedestrians walking to five rail stations to determine how far
they walked and the factors that they believed influenced their choice of route. The
survey results suggest four primary conclusions that have important implications
for designers and planners working to improve the built environment for
pedestrians.

(1) Pedestrians walk considerably farther to access rail stations than commonly assumed
Conventional wisdom among planners has often been that pedestrians in the US
will only walk a quarter to a third of a mile for any reason, including to access
transit. A paper from the mid-1990s looking at how far transit agencies and
transportation modellers assume that pedestrians will walk to a light rail stations
found very short distances, most well under a half mile (O’Sullivan & Morrall,
1996). The results of the study here suggest quite a different reality, at least for
walk trips to access rail transit. Half the people surveyed walked at least a half a
mile to access the train station (the median trip distance was 0.47 miles). This

Table 8. Attitudes towards walking

Question: For each statement below, please mark how strongly you agree or disagree with it.

Disagree or
Strongly strongly disagree
Statement agree % Agree % %

I like walking 78 19 3
Walking is relaxing 70 27 3
I walk to get exercise or other health benefits 71 23 6
I sometimes walk because it is the fastest and/ 55 34 12
or most convenient way to get somewhere
I sometimes walk because it is the cheapest 46 34 19
way to get around
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

94 A. W. Agrawal et al.

result indicates that planners and designers laying out transit-oriented

developments can assume that many train riders will be willing to walk
considerably farther to the station than they may have previously thought, at least
for commute trips.

(2) Pedestrians believe that their primary consideration in choosing a route is minimizing
time and distance
The survey explored the reasons that pedestrians choose particular routes in two
ways, first asking an open-ended question and then asking respondents to rate the
importance of a list of factors that might have influenced them. In both cases,
respondents overwhelmingly indicated that their first priority was choosing the
most direct and/or quickest route. This emphasis on convenience is probably
representative of all pedestrians, although it might also be particular to the study
respondents, who were making a morning commute trip, when time pressures are
likely to be of particular concern. This caveat notwithstanding, the results suggest
that designers and planners laying out neighbourhood street patterns should
create direct routes for pedestrians to access all major destinations.

(3) Secondary factors influencing route choice are safety and, to a lesser extent,
attractiveness of the route, sidewalk quality and the absence of long waits at traffic lights
In both the open-ended and closed-ended questions about route choice, the most
highly-rated factors after distance related to safety. In the open-ended question,
safety factors were the only other issue listed by over a quarter of respondents. In the
closed-ended questions, about half of respondents rated as ‘very important’ having
traffic devices present and traffic driving at safe speeds. The next most-cited very
important factor was having sidewalks in good condition (43%). Aesthetic issues, in
the sense of attractive landscaping or buildings, were rated as very important by
35% of respondents, but mentioned by only 8% of the respondents in the open-
ended question. These two percentages, especially the latter, indicate that
environmental appearance was considerably less of a concern than convenience
and safety. The only other issues rated as very important by at least a quarter of
respondents were having other people present (which may be a safety-related
concern), and the absence of traffic lights with a long wait.
When interpreting these results, it is important to remember the context in
which respondents answered. First, all were thinking about a commute trip in the
morning; for other trip purposes, their responses might vary. (For example, the
appearance of the built environment might be more important for leisure or
shopping trips.) In addition, audits of the neighbourhoods around the five stations,
which were conducted for this study, showed that the pedestrian environment was
relatively safe from crime and traffic, and most of the residential streets were at least
moderately attractive environments (Schlossberg et al., 2007). That is, the condition
of properties, presence of sidewalks, general tree cover and other types of urban
form elements displayed no obvious warning or danger signals for pedestrian
movement, even though conditions varied somewhat across the neighbourhoods.
Had the survey been conducted in more run-down neighbourhoods, respondents
might have placed higher priority on the visual quality and maintenance of the built
environment.
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

How Far, by Which Route and Why? 95

(4) To increase walking, focus public resources on providing direct and safe routes
Overall, the study results suggest that planners trying to encourage walking
should focus on creating direct routes and street-crossings that feel safe to
pedestrians. In fact, it may be that street crossings should be the key focus for
designers since these often represent the greatest real and psychological travel
barrier and trip delay for pedestrians.
Although improving neighbourhood aesthetics is critically important for
many reasons—such as making walking more pleasant, creating a neighbourhood
sense of place and perhaps encouraging recreational walking—these design
factors appear less critical in people’s decision of whether or not to walk for the
utilitarian purpose of getting to a rail station. Perhaps the fundamental question
that urban designers and planners wanting to promote walking should ask is not
“How can good urban design induce greater numbers of walking trips?” but
“What is the minimum standard of design that will accommodate people who
want to make utilitarian walk trips?”

Recommendations for Future Research

In addition to lessons for practicing designers and planners, this study suggests
several avenues for further research, both on how to design pedestrian surveys
and additional research topics related to the impact of the build environment on
pedestrians.
A first key finding about pedestrian survey design is that asking respondents
to trace their walking route on a local map—a relatively untested method in walk
trip research—works well. For this study, it was unsure whether respondents
would be willing to provide this information or if they would fill out the map
clearly so that the data would be useful. The study results show that the survey
technique is highly effective. Of the 328 surveys received, the map was filled out
93% of the time, and a total of 261 traced routes were usable to analyze for actual
distance and other route characteristics. These route tracings were legible and
precise enough that the research team had no trouble transferring the exact routes
into a GIS database where the distance could be automatically calculated. In
addition, the relatively high response rate for the survey (45% of distributed
surveys were returned) shows that the presence of the map did not discourage
people from completing the survey. Future researchers needing to assess the
precise distances and routes people walk should consider asking respondents to
draw their routes on a map.
A second key finding is that pedestrians vary considerably in how accurately
they estimate the distance of a regular walk trip. Many travel surveys ask
respondents to self-report the distances they travel. To date, there has been little
information on how accurate those self-reported estimates might be. This study
found that, on average, the difference between actual and perceived distance is
modest. At least half of all respondents guessed within 0.13 miles of their actual
route length. However, 25% of respondents’ guesses were off by more than 50%,
suggesting that a substantial minority do not have a precise idea of how far they
walked. A few of the individual guesses were also substantially off in terms of
distance, as well as percent: guesses ranged from up to 1.07 miles over to 0.88 miles
under the correct distance. Other surveys asking people to report the distances
about similar trip types probably have similar (in)accuracies, although the study
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

96 A. W. Agrawal et al.

results should not be assumed to hold true for self-reported distances of other trip
types. This study asked people to estimate the distance of a route they walk
routinely; people probably guess less accurately for trips they do not make
regularly. In addition, people walking significantly longer distances might
estimate distances less accurately than the respondents here, who were walking
relatively short distances.
The study also suggests new topics meriting further study. One useful
variation on the project described here would be to survey people taking different
types of trips. Future studies could target pedestrians walking to destinations
such as shopping, local services or schools to see how far they travel and whether
they similarly prioritize fast and safe routes. The study methods could also be
applied to different populations. The elderly, children and adults who do not
work outside the home might have very different route choice preferences from
morning commuters. A third useful application of the study methods would be to
research a neighbourhood with more overtly unpleasant walking conditions. The
study areas investigated were relatively safe, and while not all corridors were
exactly beautiful, there were few obvious deterrents to walking such as vacant
lots, abandoned buildings or highly dangerous intersections. Research in less
pleasant neighbourhoods could investigate whether pedestrians are willing to
shift their routes when the difference in quality between good and bad blocks is
more extreme. More detailed investigation is needed into the relative strength of
the pull of a nice environment and the push of an unpleasant one.
The need for design researchers to focus on the push of unpleasant
environmental design factors was highlighted by a related part of this research
project not covered in this paper (Schlossberg et al., 2007). In addition to surveying
pedestrians, a comprehensive audit was conducted of the physical environment
around the Japantown and the El Cerrito Plaza rail stations, and pedestrians were
also asked to indicate locations they avoided. These audits and survey questions
revealed that arterials or collectors, rather than residential streets, were the areas
respondents most often identified as poor walking environments. These findings
indicate that designers and planners should focus their attention on arterial and
collector streets. In essence, the planning and design focus for walkability in many
neighbourhoods might become, “What makes a major collector or arterial road
more or less pedestrian friendly?” Researching route choices and route avoidance
by pedestrians along high-traffic-volume streets would allow policy makers to
focus resources and interventions where they are most needed.

Conclusion
As communities wrestle with the interconnected issues of obesity, sprawl and
quality of life, this study offers rays of hope. People walked longer distances to
transit stations than previous research seems to indicate. For planners and urban
designers, this finding expands the community canvas within which to ply their
craft. In addition, the study found that pedestrians tend to value quick routes over
any other explicit characteristic of the pedestrian environment, at least for the
morning commute. This finding suggests that building basic path infrastructures
that offer direct and safe connections to popular destinations is enough to support
utilitarian walk trips, even if budgets do not permit aesthetic enhancements along
those routes.
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

How Far, by Which Route and Why? 97

Acknowledgements
The authors wish to thank the Mineta Transportation Institute at San José State
University for funding this project, Vanessa Bekkouche for helping to design the
survey instrument, and two anonymous reviewers for their helpful comments.
The contents of the paper reflect the views of the authors, who are responsible for
the facts and accuracy of the data presented.

References
Agrawal, A. W. & Schimek, P. (forthcoming) Extent and correlates of walking in the USA: an analysis of
the NHTS, Transportation Research Part D.
Badoe, D. A. & Miller, E. J. (2000) Transportation-land-use interaction: empirical findings in
North America, and their implications for modeling, Transportation Research Part D, 5(4), pp.
235–263.
Carp, F. M. (1971) Walking as a means of transportation for retired people, Gerontologist, 2(2),
pp. 104– 111.
Cervero, R. & Gorham, R. (1995) Commuting in transit versus automobile neighborhoods, Journal of the
American Planning Association, 61(2), pp. 199–219.
Cervero, R. & Kockelman, K. (1997) Travel demand and the 3Ds: density, diversity, and design,
Transportation Research Part D, 2(3), pp. 199 –219.
City Engineering Department, Vancouver, BC (May 1978) Downtown Pedestrian Surveys.
Committee on Physical Activity, Transportation Research Board (2005) Health, Transportation and Land
Use, Does the Built Environment Influence Physical Activity? Examining the Evidence. Transportation
Research Board Special Report 282.
Crane, R. (2000) The influence of urban form on travel: an interpretive review, Journal of Planning
Literature, 15(1), pp. 3 –23.
Ewing, R., Clement, O., Handy, S., Brownson, R. C. & Winston, E. (July 2005) Identifying and Measuring
Urban Design Qualities Related to Walkability. Final Report prepared for the Active Living Research
Program of the Robert Wood Johnson Foundation).
Greenwald, M. J. & Boarnet, M. G. (2001) The Built Environment as a Determinant of Walking Behaviour:
Analyzing Non-Work Pedestrian Travel in Portland, Oregon. Paper UCI-ITS-AS-WP-01-4 (Center for
Activity Systems Analysis, University of California at Irvine).
Handy, S. (1996) Urban form and pedestrian choices: study of Austin neighborhoods, Transportation
Research Record, 1552, pp. 135 –144.
Hopkinson, P. G., May, A. D. & Turney, I. G. (December 1987) The influence of town center conditions
on pedestrian trip behaviour: results from a household survey in two locations. Working Paper 246
(Leeds: Institute for Transport Studies, University of Leeds).
Hoxie, R., Rubenstein, L., Hoenig, H. & Gallagher, B. (1994) The older pedestrian, Journal of the
American Geriatrics Society, 42(4), pp. 444–450.
Kirtland, K. A., Porter, D. E., Addy, C. L., Neet, M. J., Williams, J. E., Sharpe, P. A., Neff, L. J., Kimsey, C.
D. & Ainsworth, B. E. (2003) Environmental measures of physical activity supports: Perception
versus reality, American Journal of Preventive Medicine, 24(4), pp. 323–331.
Lövemark, O. (1972) New approaches to pedestrian problems, Journal of Transport Economics and Policy,
6(1), pp. 3–9.
May, A. D., Turvey, I. G. & Hopkinson, P. G. (April 1985) Studies of pedestrian amenity. Working Paper
204 (Leeds: Institute for Transport Studies, University of Leeds).
Mitchell, C. G. B. & Stokes, R. G. F. (1982) Walking as a Mode of Transport. TRRL Laboratory Report 1064.
Nielson, G. K. & Fowler, W. K. (1972) Relation between transit ridership and walking distances in a low-
density Florida retirement area, Highway Research Record, 403, pp. 26–34.
O’Sullivan, S. & Morrall, J. (1996) Walking distances to and from light-rail transit stations, Transportation
Research Record, 1538, pp. 19 –26.
Pikora, T. J., Bull, F. C. L., Jamrozik, K., Knuiman, M., Giles-Corti, B. & Donovan, R. J. (2002)
Developing a reliable audit instrument to measure the physical environment for physical activity,
American Journal of Preventive Medicine, 23(3), pp. 187 –194.
Rodrı́guez, D. A. & Joo, J. (2004) The relationship between non-motorized mode choice and the local
physical environment, Transportation Research Part D, 9(2), pp. 151–173.
Downloaded By: [University of California Berkeley] At: 17:17 8 February 2008

98 A. W. Agrawal et al.

Schlossberg, M., Agrawal, A.W., Bekkouche, V.L. & Irvin, K. (2007) How Far, By Which Route, and Why? A
Spatial Analysis of Pedestrian Preference (San Jose: Mineta Transportation Institute, San José State
University).
Seneviratne, P. & Fraser, P. (1987) Issues related to planning for pedestrian needs in central business
districts, Transportation Research Record, 1141, pp. 7 –14.
Shriver, K. (1997) Influence of environmental design on pedestrian travel behavior in four Austin
neighborhoods, Transportation Research Record, 1587, pp. 64 –75.