Full text of DOI:10.1080/00420980124102

Urban Studies
http://usj.sagepub.com/
Encouraging Walking: The Case of Journey-to-school Trips in Compact Urban
Areas
Colin Black, Alan Collins and Martin Snell
Urban Stud 2001 38: 1121
DOI: 10.1080/00420980124102
The online version of this article can be found at:
http://usj.sagepub.com/content/38/7/1121
Published by:
http://www.sagepublications.com
On behalf of:
Urban Studies Journal Foundation
Additional services and information for Urban Studies can be found at:
Email Alerts: http://usj.sagepub.com/cgi/alerts
Subscriptions: http://usj.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://usj.sagepub.com/content/38/7/1121.refs.html
>> Version of Record - Jun 1, 2001
What is This?
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

–
Urban Studies, Vol. 38, No. 7, 1121 1141, 2001
Encouraging Walking: The Case of Journey-to-
school Trips in Compact Urban Areas
Colin Black, Alan Collins and Martin Snell
[Paper Ž rst received, September 1999; in Žnal form, July 2000]
Summary. An activity-based approach is used to analyse one speciŽ c short-trip purpose that
has persistently frustrated transport analysts trying to induce more walk, cycle or public
transport based trips. This study of the journey-to-school trip is motivated by a general consensus
that, to effect a more sustainable transport system, there is a necessity to reduce car use
(especially for short trips). Resistance to modal transfer from cars has been shown to be
embedded in various psycho-social obstacles which are not readily teased out in orthodox
econometric studies of travel demand. We report on an empirical study which fuses psychometric
(construction of coping scales) and econometric analyses (logit analysis) in an attempt to uncover
the psychological and sociological factors in uencing modal choice, as well as the usual range of
economic and demographic factors.
1. Introduction
Car ownership and use continue to rise de- (for a survey of the latter, see Goodwin,
spite a general recognition that it is neither 1992; and Oum et al., 1992). The study uses
physically possible, nor socially desirable, conventional attitudinal scaled questions,
nor economically justiŽ ed, nor environmen- although the questions themselves were de-
tally attractive, to continue to pursue policies veloped via more formal psychological pro-
that cater for and encourage greater depen- cedures than has been customary in travel
dence on road-based transport. The physical surveys hitherto. The approach attempts to
effects in terms of global pollution and local take account of the complexities of changing
environmental quality (Hughes, 1993; White- car-use behaviour. In the past, psychological
legg, 1993; Royal Commission on Environ- and sociological aspects of travel behaviour
mental Pollution (RCEP), 1995) and the have tended to be overlooked “because be-
economic costs (Peirson et al., 1996, Mad- haviour based research does not Ž t comfort-
dison et al., 1996) of such policies have been ably within the established framework of
discussed elsewhere. This paper applies a transportation modelling” (Lee-Gosselin and
variant of the activity-based approach (Jones Pas, 1997). That said, simple attitudinal mea-
et al., 1983) to consider the problems of sures have been used in earlier studies to
eliciting modal transfer from cars, rather than explore the overall bundles of modal charac-
focusing on broad generic elasticity measures teristics (comfort, security, reliability, etc.)
Colin Black is with WS Atkins Consultants, Auchinleck House, Broad Street, Birmingham, B15 1DJ, UK. Fax: 0121 643 9688. E-mail:
csblack@wsatkins.co.uk. Alan Collins and Martin Snell are in the Department of Economics, University of Portsmouth, Milton
Campus, Locksway Road, Milton, Southsea, Hants, PO4 8JF, UK. Fax: 023 9284 4037. E-mail: Alan.Collins@port.ac.uk and
Martin.Snell@port.ac.uk.
0042-0980 Print/1360-063X On-line/0D1o/0w7nl1oa1d2e1d-f2ro1m uÓsj.s2a0g0ep1ubT.chome aEtdTiRtEoNrsT UoNfIVUornbOanctoSbetur 1d6i,e2s014
DOI: 10.1080/00420980120051684

COLIN BLACK ALAN COLLINS AND MARTIN SNELL
1122
,
2. Short Trips, Mode Choice and Sustain-
ability in the Journey to School
present in private cars and public transit
(Hensher et al., 1975; Johnson, 1978; Orski,
1990).
In an activity-based framework, choice of Car-owning households make nearly double
trip time may be more in uential on house- the amount of journeys by motorised means
hold decision-making on a day-to-day basis than a household without a car, and many of
than mode choice. Indeed Jones et al. (1983) the journeys travelled by car are short dis-
shows that mode choice is typically deter- tances previously made on foot, or public
mined by scheduling considerations, such as transport or by cycling (Wooton, 1993). One
the need to see children to school by a certain of the most obvious ways in which car use
time and to arrive at work by a certain time. will have to change if transport is to become
This approach throws into sharper relief the more environmentally benign, is in its use for
interdependencies and linkages neglected in journeys of a short distance. A signiŽ cant
traditional transport modelling such as multi- and rapidly growing proportion of car use is
purpose trips, driver- and passenger-trip pur- for very short journeys, to destinations that
poses and joint schedule accommodation. are within practical walking or cycling dis-
Accordingly, the design of sustainable trans- tance (Goodwin, 1995). Often, these jour-
port policy needs to take cognisance of the neys will be between 2 and 5 minutes in
speciŽ c behavioural complexities associated duration. When allowance is made for start-
with a wide range of particular journey pur- ing up, travelling in urban areas and parking,
poses if it is to be successful in eliciting the time by car for these journeys will often
modal transfer away from cars. This study not be vastly different from that if the jour-
considers one particular short-trip purpose— ney were by foot. This pattern of car use is
the journey to infant school in a compact not conŽ ned to Britain. Ullrich (1990) found
urban area. This journey type has been that in Germany (former Federal Republic)
recognised as a pressing concern in the de- approximately 3 per cent of all journeys were
sign of sustainable urban transport policies no greater than 500 m. One-third of all car
(Department of the Environment, Transport journeys were no greater than 3 km and one-
and the Regions, 1998). At Ž rst sight, this half no more than 5 km.
journey type appears to be one of the most
Short trips do not represent a large pro-
promising for encouraging car trip reduction, portion of total mileage (1 mile 1.6 km),
5
given the high density of infant schools in but they are important for a number of rea-
English towns and cities such that a short- sons. Many of these journeys will be dispro-
walk trip is a realistic option. A majority of portionately
inefŽ cient
in
energy
children attending infant school in England consumption when compared to longer jour-
generally live within 2 km of the school and neys, because of the inefŽ ciencies associated
many live much closer.
with driving a cold engine, on local roads
The paper is organised as follows. In sec- and in urban trafŽ c conditions (Potter and
tion 2, the factors in uencing mode choice in Hughes, 1990; RCEP, 1995). On short-dis-
the journey to school, in the UK generally tance journeys, the engine is cold and runs
and in our survey areas speciŽ cally, are dis- least efŽ ciently. Those cars Ž tted with cata-
cussed in the context of sustainable transport lytic converters usually do not have sufŽ cient
policy. Preliminary model development time to reach operating temperature for the
based on conventional economic and demo- catalytic converter actually to work. Accord-
graphic factors is presented in section 3. ingly, encouraging walking and cycle trips
Further model development that also incor- has been the focus of much research attention
porates the psychometric analyses is set out (Tolley, 1997; Atash, 1994; Cervero and
in section 4. The results are discussed in Radisch, 1996).
section 5. Concluding remarks are offered in
the Ž nal section.
The growth of very short car journeys is
particularly prevalent in the use of the car for
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

ENCOURAGING WALKING
1123
–
Children aged 5 10 years: journeys to/from
Table 1.
–
school by main mode 1975/76 and 1989 94
1975/76
(percentage)
1989/94
(percentage)
Mode
Walk
Cycle
Car
Bus
Other
71.7
1.3
15.6
10.8
0.6
62.3
0.7
28.2
7.5
1.3
Source: Department of Transport (1995).
escort trips, especially for education, which from each respondent on how their child(ren)
have grown twice as fast as the average use usually gets to school, according to the age
(Goodwin, 1995). Short journeys by parents of the child attending (eldest if more than 1).
driving their children to school contribute A total of 34 respondents did not answer the
approximately 20 per cent to all UK car mode-of-travel question and have been ex-
journeys on weekday rush hour mornings. cluded from all analyses and tables. Table 2
More than 50 per cent of these trips cover suggests that the sample is reasonably spread
distances under one mile, and the number of across all age-groups at infant level. More
children travelling to school by car has dou- importantly, it shows that over 96 per cent of
bled over the past 20 years to 1 in 4. Table 1 children of this age travel to school either on
indicates the predominant shift in the mode foot or by the household car. This is perhaps
used to travel to school by children of aged not surprising given the urban nature of the
–
–
5 10 years over the period 1975 94. Over sample, where distances to school are gener-
this period, the proportion of school journeys ally short. The overall proportions are very
made on foot has decreased by 9.4 percent- comparable to those in Table 1, although a
age points and bus use has fallen by 3.3 lower percentage in our sample go by bus,
points. Car use has increased by 12.6 per- re ecting the lower mean age considered.
–
centage points.
Over the period 1975 94 there were
In order to investigate factors behind mode signiŽ cant changes in the distance that chil-
choice for the journey-to-school trip, a self- dren travelled by each mode in the UK.
completion questionnaire survey was con- Children walking over half a mile to school
ducted in 51 infant schools or combined have fallen in numbers by 13 per cent. Only
–
infant/primary schools (schools with pupils 24 per cent of children age 5 10 undertook a
up to age 11) in south-east Hampshire (36 journey of more than half a mile to school.
schools) and central north-west England (15 Conversely, car journeys to school for this
schools). Questionnaires were issued to age-group have increased in number, but de-
schools to be completed by the person most creased in distance. There has been a 4.5 per
responsible for ensuring their child reached cent increase in the number of school jour-
infant school each day. These were returned neys by car less than a mile over the period
–
to the child’s teacher. The survey took place 1975 94.
over May and June 1996. In all, 4214 re-
Table 3 provides comparable evidence
sponses were received, of which 3333 were from the current survey. Respondents were
from the south-east Hampshire area and 881 asked how far it was to the infant school
from the central north-west England (Stoke- from home in terms of the approximate num-
on-Trent/Crewe) area. The overall response ber of miles. The answers thus refer to per-
rate was approximately 36 per cent.
ceived distances, although since in practice a
Table 2 provides the summary information large number stated the estimated distance to
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

COLIN BLACK ALAN COLLINS AND MARTIN SNELL
1124
,
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

ENCOURAGING WALKING
1125
Percentage who travel by each mode
Distance to school from home (miles)
Table 3.
–
–
, 0.5
5
0.5 1.0
5
1.01 2.0
5
. 2
5
Mode
(n 1113)
(n 1743)
(n 408)
(n 238)
Walk
89.5
0.3
9.3
0.6
0
0.3
0
66.4
1.2
30.1
2.0
0.3
0
27.7
2.0
62.7
4.9
0.7
1.0
5.5
0.8
81.9
3.8
3.4
2.5
Cycle
Own car
Friend’s car
Public transport
School bus
Taxi
0
1.0
2.1
Total
100.0
100.0
100.0
100.0
Note: Question is ‘How far is it TO this infant school from home? …… miles
approx.’.
the nearest quarter of a mile, they are likely that feature in the environment of many chil-
to be more than adequately accurate for the dren (Stokes and Hallet, 1992). The pro-
shorter journeys. The mean travel distance motion of these car-focused social norms
over the 3529 respondents who provided this may thus be countered by addressing
information is 0.92 miles but it is quite clear parental decisions. This is evident in research
that distance to school is a crucial factor in by Hillman et al. (1990) who report on
determining mode of transport. The further changes in parents’ perceptions of the safety
the distance, the more likely that a car is of their children, and their impact on the
used. However, this is obviously not the only personal independent mobility of children.
determinant since for almost any distance They show that, on average, today’s parents
chosen there are some children who walk to will not allow their child to cycle to school
school. In order to promote less use of cars alone until the age of 12, although they
for school-trip purposes, it is necessary to observed differences between bicycle-own-
isolate those other factors that determine ing and non-bicycle-owning parents (the for-
mode choice. This is attempted in the next mer allowing cycling to school alone at an
section.
earlier age). James (1995) argues that chil-
The trends noted above are signiŽ cant for dren who are not allowed to walk or cycle to
trafŽ c and pollution levels from a local sus- school are being conditioned to a life of
tainability perspective where ‘liveability’ is inactivity, which will have serious implica-
important. Further, they also have some other tions for their susceptibility to obesity and
worrying implications, since young children coronary heart disease. Further, Armstrong
become accustomed to this pattern of car use. (1993) suggests that patterns of lifelong
This has the potential to communicate unsus- physical activity are most likely to be started
tainable values to children from an early age, in childhood.
which may become increasingly difŽ cult to
Hillman et al. (1990) also suggest some
change as they get older, particularly since unintended consequences of government pol-
cognitive rules are already well developed icy that are problematic. They argue that the
(Petty and Cacioppo, 1986). In this way, car nature of the government road safety cam-
use and car-dependent lifestyles become es- paigns has led to a loss in the level of child
tablished at very young ages, assisted by the independent mobility. They argue that the
psychological reinforcement presented by fear these campaigns have instilled has
numerous media and cultural images of cars prompted a steady increase in the amount of
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

COLIN BLACK ALAN COLLINS AND MARTIN SNELL
1126
,
Likely overriding factors
Table 4.
Number of
respondents
answering
Percentage
of valid
responses
in category
this question
(maximum 4023)
Category
Not holding a driving licence
With no cars in the household
Never having a car available to
take the children to school
21.2
13.8
23.6
4008
4011
3906
time parents spend escorting their children to subsequent analysis concentrates on this two-
school. For many parents, they argue that this way split and, unless speciŽ ed, excludes
has induced a greater reliance on the motor households using the other modes of travel.
vehicle and has had unintended conse-
A number of factors are likely to provide
quences in terms of trafŽ c congestion, danger overriding constraints on choice of travel
and pollution.
mode—for example, where there are no cars
reported in the household, or where the re-
spondent does not possess a driving licence.
Table 4 indicates the importance of these
criteria. Since there would nearly be 100 per
cent predictive power of travel mode choice
for a large number of households if these
factors were incorporated, all cases where at
least 1 of these 3 questions was answered as
in the table have been excluded. This re-
duced the sample by 1288 respondents from
the 4023 of Table 2 (walk or own car) down
to 2735.
A range of other factors were also in-
cluded in the questionnaire and basic statis-
tics for those referred to in this paper are
summarised in Table 5. The percentage of
respondents who use a car increases to 42.2
per cent, naturally re ecting the exclusion
from the sample of those without the avail-
ability of a car, etc. Whilst the Ž nal column
might be indicative of factors in uencing
mode of travel, this does not control for all
the factors together.
3. Modal Choice for the Journey-to-school
Trip: Economic and Demographic Factors
Choice of transport mode is strictly an inte-
gral part of the household decision-making
process. As such, it is a complex issue which
can only be handled by a very large-scale
interconnected model. For example, whether
a car is available at all depends on the house-
hold’s decision on car ownership, which may
be linked to income and household employ-
ment decisions. Further, one could envisage
a simultaneous model where household resi-
dential location and car ownership were the
crucial factors, these being dependent on em-
ployment availability, size of household and
the location of schools, right up to secondary
level. Estimation of such a model is beyond
the scope of this paper and we therefore
make the simpliŽ cation that certain factors
are exogenous to choice of travel mode.
Although seven modes of travel were elic-
ited from the questionnaire, Table 2 showed
Employment type is important for several
that the choice is effectively between walk reasons. It acts, alongside the car ownership
and use own car. In addition, the use of variable, as a proxy for income. This ar-
public transport, taxis, friends’ cars, etc. is guably may both in uence attitudes to the
likely to be decided by a very large number environment and also re ect marginal time
of factors (such as location of bus-stops, valuations. Over 75 per cent of the sample
income of friends) of which we have no were either full-time housepersons or were
knowledge. To avoid biasing any results, all employed on a part-time basis, re ecting the
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

ENCOURAGING WALKING
1127
Summary of variables used
Type of
Table 5.
Percentage
of
Percentage
of category
using car
Variable
name
DeŽ nition
variablea respondents
MODE
AG2
Usually uses car
D
D
D
D
D
D
D
C
D
C
D
D
42.2
3.5
100.0
45.3
40.5
38.8
47.6
47.4
40.6
6.00b
43.9
1.54b
47.3
40.5
Age-group , 26 years
–
AG3
Age-group 26 30 years
22.1
41.9
23.4
9.2
–
AG4
AG5
AG6
SEXD
AGECA
TWOKIDS
DIST
COMPCAR
PARKD
Age-group 31 35 years
–
Age-group 36 40 years
Age-group 41 years
1
Sex: male
7.4
Age of Ž rst child in years
Has 2 children at infant school
Distance in miles
Has company car in household
Considers there to be a parking
problem at the school
Southern county
5.99b
24.3
0.97b
18.2
74.0
AREA
D
D
D
D
D
D
D
76.5
58.7
15.4
45.3
2.6
30.9
5.8
36.2
33.9
47.2
44.4
44.3
35.0
45.6
ONLY1CAR Has just 1 car in household
E1
E2
E3
E4
E5
Employment type: full-time
Employment type: part-time
Employment type: shift-work
Employment type: f/t houseperson
Employment type: not in employment
a D dummy; C continuous.
5
5
b Statistic refers to mean value over sample.
likelihood that the person responsible for the model,1 several simpliŽ cations were con-
journey to school was not usually the main sidered
household earner. Accordingly, for the great
(1) Assume that mode choice was largely
majority of survey respondents, it would be
unrelated to journey purpose, after con-
reasonable to infer that their time constraints
trolling for employment;
are unlikely to be binding. For those who
(2) Exclude those who made this sort of
work and use a car to reach their place of
multipurpose trip, on the grounds that
employment, the choice of mode for school
this is only a relatively small percentage
trips becomes much more complex, involv-
of the sample, although recognising that
ing the location of the work-place and simi-
we would lose the information provided
larity between work and school start-times.
by any link between school trips and
The whole process could be endogenous
journey-to-work trips and possibly bias
within the household so that any household
the overall results.
earner (the main earner or any other) chose
(3) A Ž nal possibility was to consider a
their work location to Ž t around school trips.
three-way modal choice—walk, special
Several of the questions dealt with
trip by car, or as part of a journey-to-
whether the journey to school was usually
work trip. This is strictly a nested model
part of a multipurpose trip. Of the restricted
but, given the lack of information about
sample, 16 per cent stated that the trip to
households and work locations, it could
infant school was usually part of a longer
not be set up in this manner.
journey to work, amounting to 37 per cent of
those who regularly used a car. To avoid the None of the approaches is ideal but, given
complex data requirements of a joint-choice the absence of a clear-cut rationale for pre-
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

COLIN BLACK ALAN COLLINS AND MARTIN SNELL
1128
,
Logit results: journey-to-school trips (N 2203)
Table 6.
5
Variable
CoefŽ cient
Standard error
P[uZu z]
5
Constant
DIST
AG3
AG4
AG5
AG6
E1
E3
E4
E5
COMPCAR
ONLY1CAR
AREA
PARKD
DISTSQ
SEXD
0.1290
2.4179
0.1454
0.3257
0.0582
0.0678
0.2514
0.1811
0.5321
0.1880
0.1335
0.6663
0.7852
0.5391
0.1609
0.3490
0.0374
0.0237
0.4647
0.1386
0.3070
0.2974
0.3083
0.3359
0.1712
0.3258
0.1278
0.2376
0.1463
0.1156
0.1261
0.1190
0.0240
0.2238
0.1388
0.0526
0.7813
0.0000**
0.6357
0.2735
0.8502
0.8400
0.1421
0.5784
0.0000**
0.4288
0.3615
0.0000**
0.0000**
0.0000**
0.0000**
0.1189
0.7874
0.6527
2
2
2
2
2
2
2
2
2
2
2
2
2
2
TWOKIDS
AGECA
2
Percentage of accurate predictions
76.2
** indicates signiŽ cant at the 1 per cent level.
ferring one to another, subsequent economet- school trips or may just be due to more
ric work reported here is based on option general differences in trafŽ c congestion. (No
(1).2 All model estimation was based on the information was available on comparative
standard logit formulation:
journey times between localities.) Respon-
dents who perceived there to be a parking
problem at the school were much less likely
to go by car. This may be as expected and
suggests that a policy of making parking
more difŽ cult would reduce the number of
cars used on school journeys. An interesting
issue concerns the path to any equilibrium
since, as fewer cars drive to school, parking
problems may decline, so increasing use
again.
eb x
1
9
i
Pr (mode choice is car)
5
b x
9
i
1
e
xi
where,
ables.
is the vector of explanatory vari-
Table 6 clearly conŽ rms the initial im-
pression from Table 3 that perceived distance
is a key determinant of the decision to use a
car. By holding all other variables constant at
a central value—the mode for dummy vari-
The employment variable requires careful
ables and the mean for others—the relation- interpretation. The base-line case (dummy
ship between the probability of using a car variable excluded) was for part-time employ-
and distance can be derived. The graph of ment. The results show that full-time
this is a standard ogive shape, increasing housepersons are signiŽ cantly different from
from about 20 per cent at a half a mile all other types of employment, in that they
distance, rising to 50 per cent at 1.25 miles are much less likely to use a car (suggesting
and 80 per cent at around 2 miles. Thereafter, that time spent walking has a lower oppor-
the probability gradually approaches 1. Also tunity cost for this group of persons). When
seen as signiŽ cant in Table 6 are whether the cases where the child is dropped off on the
household only had one car and the region in way to work are excluded, then it would
which the respondents lived. This may re ect appear that those in full-time employment
different attitudes to the use of vehicles for are much less likely to use a car. This is more
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

ENCOURAGING WALKING
1129
an artefact of reducing the sample by this response to a number of transport-related
means than anything else. It may also partly issues. Reser has suggested that the car has
be explained by the opposite effect, related to become a coping mechanism since it
being unemployed.
allows for some degree of adaption to a
stressful environment, both in terms of
social and physical stressors and perceived
social control (Reser, 1980, p. 286).
4. Factors behind Modal Choice for the
Journey-to-school Trip: Further Develop-
ment of the Model
However, he concedes that this individual
control may be more psychological than ob-
jective. The approach used to identify coping
mechanisms does not rely on an individual’s
The study by Bradshaw (1995) summarises
the principal factors that in uence a parent’s
decision to drive their child to school:
—length of journey to school; self-reporting of their psychological behav-
—the possibility of linking the school jour- iour. This is an attempt to avoid respondent
ney with the journey to work;
—road safety concerns;
bias, or explanations that mask the underly-
ing true mechanisms through which people
—concerns about the personal safety of chil- cogitate transport decisions. According to
dren;
Breznitz (1986, p. 326), some “behaviours
are too automatic, and probably do not take
the route of premeditation that can be re-
ported”. Houston (1987, p. 385) asserts that
“people may employ coping responses with
or without being aware of doing so”. Further
in this vein, Hallman and Wandersman
(1992) argue that people do not respond well
to spontaneous introspection about their per-
sonal coping strategies, such that the
identiŽ ed coping strategies may be inaccur-
ate. Psychometric techniques, such as those
outlined below, are considered by main-
stream psychologists to be the most suitable
means of investigating the utilisation of dif-
ferent mechanisms of coping as they are
argued to permit accurate, objective and
communicable descriptions (Guilford, 1954)
that
—parental choice of school;
—peer group pressure.
One might also include another factor—the
weather. Yet, in their research on the effects
of weather conditions upon travel decisions,
Khattak and de Palma (1997) found
signiŽ cant evidence of behavioural inertia.
They suggest that people would rather stick
to habitual patterns of travel than respond to
variable factors such as the weather.
A key aspect of our particular survey was
to move beyond the basic economic and
demographic determinants of modal choice
and to consider a wider range of possible
behavioural and psychological factors. Such
factors may be measurable and systemati-
cally linked. This aspect of the questionnaire
was developed through a number of pilot
studies. An initial pilot study of parents of
children in one infant school encouraged 187
respondents to explain in detail and with free
responses their understanding of, and atti-
tudes to, trafŽ c and transport problems.
seem well suited for identifying similari-
ties and differences among groups with
regard to risk perceptions and attitudes
(Slovic et al., 1982, p. 85).
Together with the development of a psycho- An essential step in the construction of an
logical ‘coping scale’ to address perceived attitude scale relates to ascertaining state-
threats in their transport and environmental ments that will represent in a particular test
situation, the resultant questionnaire was the universe of interest (Edwards, 1957). The
reŽ ned further through extensive piloting various informal criteria suggested by Wang
with a number of other schools. The coping (1932), Thurstone and Chave (1929), Likert
scale was devised to highlight behavioural (1967), Bird (1940) and Edwards and Kil-
coping mechanisms that may be deployed in patrick (1948) were used as a guideline in the
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

COLIN BLACK ALAN COLLINS AND MARTIN SNELL
1130
,
construction of statements. For example, most respondents disagree that governments
avoiding statements that are overly factual; will act to solve trafŽ c problems long before
avoiding statements likely to be endorsed by they get out of hand and most agree that the
everyone or no one; keeping the language majority of parents are worried that small
relatively simple, clear and direct. As in ear- children walking to school alone would be at
lier applications in other areas using coping risk from strangers.
scale development (see, for example,
The origin of the statements was described
Hallsworth, 1987), the attitudinal statements above, the variety of them encompassing a
were cast at the ‘mundane level’ at which the wide spectrum of viewpoints, both positive
debate on transport problems is conducted. and negative, towards road use. Some ques-
Indeed, some phrases were adapted from tions might be expected to elicit an ‘agree’
those previously used by politicians, and oth- viewpoint from the pro-car individual and
ers were taken from the press. Acquiescence some a ‘disagree’ viewpoint. The third
bias (the tendency for some respondents to column of Table 7 suggests a priori whether
agree with whatever they are asked in a agreement with the statement might be ex-
survey) was addressed as far as practically pected to lead to higher car use (shown as
1
possible by wording statements in different
directions (Ray, 1989).
) or lower car use (-). A question mark (?)
indicates a debatable response.3
A Likert type of attitude scale (Likert,
To examine the effect of attitudes on the
1967) was developed, whereby values are use of the car for the journey-to-school trip,
assigned along an attitude continuum from two approaches are possible. First, if there
one extreme to another. The transport coping are some distinguishable and measurable
scale used a seven-point semantic differential strands of opinion which lie behind the an-
scale of measurement, which is common to swers to the 31 statements, then the scores on
many studies of risk perception (Slovic et al., these dimensions could be added to the logit
1982). A semantic differentiation technique model. Alternatively, since there are no prob-
enables the essential operation of measure- lems with degrees of freedom, there is no
ment to be performed by the successive allo- need for any data reduction and the responses
cation of a concept to a scale deŽ ned by to all the questions can be included together
polar adjectives (Osgood et al., 1957). The as independent variables. With this approach,
judging group, or respondents, were asked to answers to individual questions might be
respond to the statement in terms of their seen as playing a highly signiŽ cant role in
degree of agreement or disagreement along explaining travel choice. The difŽ culty here
the continuum (Edwards, 1957).
may be of spurious signiŽ cance occurring
A total of 31 statements were put to re- when a large number of variables are in-
spondents to elicit their views on such mat- cluded, perhaps with no clear theoretical ba-
ters. For each of these, respondents were sis. Also, a few individuals with strong
asked to circle the number from 1 to 7 which opinions on some questions may bias the
most closely resembled their viewpoint results. Nonetheless, the approach can prove
(1 strongly agree, 7 strongly disagree). enlightening. Both approaches are followed
5
5
Table 7 presents the summary statistics for below and, since they are not distinct, a
these questions, with question numbers indi- combined model is also presented, to obtain
cating the order of their presentation. Since a the advantages of each.
score of 4 represents ‘neither agree nor dis-
agree’, this has been subtracted from the
mean to aid interpretation. Thus a positive
4.1 Factor Analysis
result shows an overall tendency for the sam- A factor analysis was performed on the 31
ple to disagree with the statement, a negative attitude questions of Table 7, using the full
result the opposite. Questions 28 and 29 are sample available. The scree plot of eigenval-
the ones showing consensus most clearly— ues suggested between 3 and 7 factors (the
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

ENCOURAGING WALKING
1131
Statements presented to respondents
Table 7.
Expected
link to
car use
Question
number Statement
Mean Standard
(
4) deviation
2
26
27
28
If I don’t use the roads—someone else will
Having a car is part of having a good lifestyle
Most British parents are worried that small children would
be at risk from strangers if they were allowed to walk to
school alone
0.46
0.11
2.10
2.02
1.94
1.57
1
1
1
2
29
30
The government will act long before trafŽ c problems get
out of hand
1.70
1.23
1.55
1.68
1
1
People spend too much time worrying about
environmental pollution
32
33
Pollution caused by cars is not a problem where I live
Protest is pointless, you will never get a government to
change its policy—unless it WANTS to
Any worries parents may have about the environment take
second place to getting their child to school safely—and
that means in the car if they have one
Nothing I do makes any difference to the environment one
way or another
Given all the air pollution in the world, a short car trip to
school makes no difference at all
It is the Government’s responsibility to worry about the
environment: not mine
No-one would give up their car just because someone
else was becoming worried about air pollution
I am far too busy to worry about environmental issues
Many people could not work if they did not have a car
Many people have to use a car everyday—even if they
would rather not
0.92
0.64
1.68
1.96
1
1
2
2
34
1.24
1.18
1.22
1.62
0.97
1.66
1.70
1.65
1.58
1.70
1
1
1
1
1
35
37
38
39
2
41
42
46
1.16
1.25
1.43
1.56
1.87
1.70
1
1
1
2
2
48
54
The motorist always gets the blame for something—air
pollution just happens to be the latest fad
Car engines can be designed that will totally eliminate
pollution
Nothing would make me give up my car
The world has far more cars than is sensible
It is morally acceptable to commit a small crime on a
road construction site in order to protest against the
greater crime of environmental destruction
Before long, our health is really going to suffer from
air pollution
I feel a moral duty to minimise pollution to preserve the
planet from damage
This country should take the lead on environmental issues
I expect that at some future date this country will have
severe pollution problems
0.56
0.98
1.71
1.59
1
1
2
2
56
31
36
0.27
1.00
1.06
1.94
1.76
1.85
1
—
—
40
43
—
—
1.40
0.68
1.60
1.48
2
2
45
47
—
—
1.33
1.57
1.44
1.41
2
2
52
If my job caused environmental problems, I’d rather be
unemployed than carry on causing pollution
Problems with car fumes are getting out of control
A serious car accident could happen at any time—it is
just another everyday risk
—
0.89
1.77
53
44
—
?
0.72
1.22
1.53
1.70
2
2
49
50
51
55
If road trafŽ c is to be cut, then company cars must be
the Ž rst to go
?
?
?
?
0.03
0.14
1.24
1.55
1.80
1.65
1.56
1.55
2
2
Only the green activists keep society alert to pollution
issues
People over-estimate risks from trafŽ c—most people
survive car accidents
There are so many cars now; you could be hurt in an
accident tomorrow
2
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

COLIN BLACK ALAN COLLINS AND MARTIN SNELL
1132
,
Ž rst 7 eigenvalues were 4.820, 2.751, 1.921, some of unclear sign. This is more than an
1.271, 1.107, 1.092 and 0.997; corresponding artefact of the data, although it might be
cumulative percentages of variation ex- argued with hindsight that more of the ques-
plained were 15.5, 24.4, 30.6, 34.7, 38.3, tions should have been stated in a negative
41.8 and 45.0). Since choice of the number stance to respondents to ensure that the re-
of factors is not an exact science, a heuristic sults were robust.
approach was taken whereby some coherence
Scores on each of these factors were de-
and intuitive rationale for the calculated fac- rived by summation of the scores for each
tors was looked for. Varimax rotation after a question in a factor, using the factor loadings
principal components extraction of factors as weights. Since the responses to each ques-
was used for the derivation and provided a tion had varying means and standard devia-
clear allocation of questions to single factors. tions, each question was Ž rst standardised to
Table 8 shows the loadings of questions for a unit mean and variance using the values of
three-factor analysis and indicates the alloca- Table 7. This ensured that the loadings to
tion when there are 4, 5, 6 and 7 factors used, each question determined its weight in the
annotating loadings under 0.3 in brackets.
factor, rather than the tendency for the ques-
The split into three factors appears to have tion to have a high or low score, etc. The
some intuitive order. It is possible to classify logit results after including the three factors
the factors as: individual responsibility and are presented in Table 9(a).
impact; environmental awareness; and car-
centredness. The Ž rst factor seems to relate
the level of intrinsic social and moral con-
4.2 Critical Question Analysis
cern about one’s individual role in active One alternative approach is simply to add the
citizenship for the community or stewardship responses to these questions directly as ex-
for the environment. The second relates to planatory variables of the model, initially all
the inate mental prioritisation of broad en- together, and then reduced because of poss-
vironmental issues in one’s lifestyle. The last ible multicollinearities to a small core of
factor clearly pertains to an individual’s level signiŽ cant questions. This Ž nal logit is pre-
of integration and dependence on car-based sented in Table 9(b). Only questions 27 and
resources in the course of their day-to-day 39 do not have the correct expected sign.4 In
lives.
terms of predicting the correct mode of
There is some overlap (for example, ques- travel, there is some difference between these
tion 43 would be expected to be in factor 1 two models—the use of individual questions
rather than factor 2) and some of the ques- predicting 79.0 per cent of correct responses
tions with lower loadings do not seem to Ž t. compared to the 77.2 per cent using factors.
However, increasing to four factors does lit- The difference in this direction is largely to
tle more than split factor 3 into two—factors be expected since using factors loses infor-
1 and 2 remain largely unchanged. In broad mation. However, both are clear improve-
terms, the same effect occurs when higher ments on the 76.2 per cent obtained when
numbers of factors are obtained. Examin- attitudes were not considered. By combining
ation of the questions in the extra factors the highest two age-groups into one, the esti-
produced did not consistently provide any mation using individual questions suggests
1
readily identiŽ able attributes. For example, that those in the 40 age range are
the questions on car accidents (Q51, Q55 and signiŽ cantly (5 per cent level) more likely to
Q44) rarely occurred in the same factor or, if use the car. This was not found when factors
they did, the logic of other factors was dis- were used. It is possible that there is a link
placed. For this reason, further analysis con- between some attitudes and age which is
siders only three factors. In terms of handled well by considering individual ques-
expected signs, factor 2 generally corre- tions.
sponds to the positive effects of Table 7, plus
Given that some questions looked out of
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

ENCOURAGING WALKING
1133
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

COLIN BLACK ALAN COLLINS AND MARTIN SNELL
1134
,
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

ENCOURAGING WALKING
1135
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

COLIN BLACK ALAN COLLINS AND MARTIN SNELL
1136
,
place in each factor, an alternative was to be problem at the school and lives in a southern
a little more selective about questions enter- county. Since each factor contains a different
ing each factor, paying attention also to the number of questions, interpretation of factor
loadings. With this in mind, questions 49, 36, scores is difŽ cult. The results for each factor
–
52, 39 and 28 (all with loadings less than 0.4) are displayed in Figure 1, parts (a) (c). Be-
were dropped from factors 2 and 3 (new cause each question is standardised to mean
factors are called FACT2R and FACT3R). zero, the mean of each factor is also zero (or
Table 9(c) shows the results of this, also approximately so). The variance of each fac-
including the questions omitted from the fac- tor differs and the diagrams therefore include
tors on their own, since they may re ect the Ž rst and third quartiles, so that the spread
special unclassiŽ able features. All three fac- of values for these factors from the sample
tors are now signiŽ cant determinants of mo- can be discerned. To obtain some idea of
dal choice. The signiŽ cance of question 28 how the responses to each question relate to
picks up the relevance of child safety in the the factor scores, the graphs are repeated
journey-to-school decision. This did not ap- with a different view of the horizontal scale
–
pear strongly in the factor analysis, unless a Figure 1(d) (f). Taking the factor 1 graphs,
factor in its own right. Although child safety one can imagine an individual who scores 1
might not seem a crucial factor, given that in (strongly agree) with each of the 11 ques-
this sample few parents permitted their child tions making up this factor. After standardis-
to walk to school unaccompanied, the car ation, this is equivalent to a score of
13.5
2
may be chosen where the opportunity cost of for that factor and must be the minimum
walking with the child is high. As a Ž nal test score for that factor. Similarly, responding
to show the combined relevance of the attitu- with all 7s leads to the maximum score of
dinal variables taken as a whole, a Wald test 6.4. The values on the horizontal scale ex-
of restricting the coefŽ cients of these vari- tend this idea to intermediate values and
ables (the last 8 in Table 9(c)) to zero pro- represent probabilities for respondents who
vided a highly signiŽ cant chi-squared value answered identically to each question in the
(22 degrees of freedom) of 64.4.
factor. Obviously this is unlikely for any
particular individual, but it does aid interpret-
ation of each factor.
5. Analysis of Results
To assess the relative effectiveness on re-
The regressions above show that car use on ducing car use of changes in each factor,
journey-to-school trips can be reduced by a consider the required change in factor score
policy which affects parents’ mean scores on necessary to reduce the probability of car
any of the three factors. At the margin, how- usage by 5 percentage points from its value
ever, differences in elasticities may suggest at mean factor score (0.27)—i.e. down to
that it is preferable to concentrate on one 0.22. For factor 1, this requires an increase
factor rather than another. In order to assess in the score from
0.07 to 4.00, a total
2
the extent to which changes in parental atti- change of 4.07; for factor 2, the score must
tudes, as captured by the three factors, might change from 0.45 to 0.49—i.e. by a total
2
lead to shifts in travel mode for the journey- amount of 4.94; for factor 3, the total change
to-school trip, simulations were performed required is from 0.40 to 1.10, equal to
2
using the model of Table 9(c). All other 1.50. There is some difŽ culty in interpret-
variables were held constant at a central ation of these results since each factor is
value—the mode for dummy variables and measuring a different characteristic and
the mean for all others. This results in a could be measured in different units. Thus
typical-case scenario whose major character- the two changes required cannot apparently
–
istics are: age 31 35, female, works part- be directly comparable. However, it would
time, household owns only one car (not a appear that to obtain the desired change in
company car), considers there to be a parking car usage by attempting to amend awareness
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

ENCOURAGING WALKING
1137
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

COLIN BLACK ALAN COLLINS AND MARTIN SNELL
1138
,
of the environment alone is the most changes in score required to reduce the prob-
difŽ cult. ability of using a car by 5 per cent are now
Figure 1, parts (d), (e) and (f) provides standardised by the standard deviation of the
another angle to the question. Since the lines factor. This provides Ž gures of 1.37 standard
are roughly straight (although (f) shows signs deviations for factor 1, 2.01 standard devia-
of  attening with the high scores), a com- tions for factor 2 and 0.88 standard devia-
parison of the slopes may suggest whether tions for factor 3. Thus, the large standard
one factor may be more effective as a means deviation of factor 1 to some extent counter-
of reducing car use. Starting at a situation acts the relatively large change in score re-
where a typical person scored 4 on each quired. Changing attitudes to individual
question (the exact starting-point is not cru- responsibility and impact does still, however,
cial), the score would need to change by appear less useful than acting on car-centred-
–
1
1
1.12, by 1.57 and by 1.07 on factors 1, ness. The environmental awareness factor
2 and 3 respectively to decrease car usage by again seems less in uential as a policy focus.
5 percentage points (assuming the score on Of course, these results must be tentative.
the other factors remained the same). This Further research is desirable into whether it
repeats the previous Ž nding that changes in is easier to in uence opinion on some factors
environmental awareness have less impact rather than others. It is also worth noting that
than changing other attitudes. An across-the- these results necessarily assume that the rela-
board change to responses on the car-cen- tionships between factor scores and car use
tredness factor would seem to produce the are stable. It is quite possible that changes in
greatest change in car usage. The shape of elasticities can be induced by environmental
this relationship over the relevant range ‘scares’ over the effects of pollution or con-
clearly reveals that the gains are easier to gestion, campaigns on road safety, etc.
obtain when car-centredness is already high. SpeciŽ c experiments would need to be de-
Clearly, it may be easier to change opin- signed to assess their in uence.
ions on one factor than another. A full exam-
ination of this is beyond the scope of this
paper, although must be considered before
6. Concluding Remarks
any policy implementation. However, some The journey to school has gained the atten-
tentative pointers can be made by examining tion of those advocating a more sustainable
the spread of responses for each factor. Al- transport system, probably because it is eas-
though the (imaginary) units in which factor ily identiŽ able and gives the impression of
scores are measured are not on a ratio scale, being a relatively easy journey to target for
there are clear minimum and maximum car-trip reduction. The existence of a number
scores. It is a possibility that, where there is of psycho-social obstacles and intrahouse-
a wide distribution of scores between these hold trip interdependencies, however, renders
points, it will be easier to change scores in even this trip purpose much more complex to
absolute terms than where there is a very address than might appear at Ž rst glance.
concentrated distribution. For example, the Following an extensive survey of parents
range for factor 1 is large compared to factor with children of infant school age, be-
3—would it be intuitively reasonable to ex- havioural determinants of travel mode choice
pect the same change in score on each factor decisions were teased out via a fusion of
to lead to the same change in probability of psychometric and econometric analyses, in-
car use? If it is believed that a relatively corporating standard demographic variables
larger range/spread of the distribution does and travel distances. Regarding travel dis-
make it easier to effect a change, then we can tances, it seems that promotion of walking
incorporate the standard deviations to com- has greater prospect if school catchment is
pare effectiveness. Standard deviations on explicitly considered. Hence, such an ap-
each factor are 2.99, 2.46 and 1.71. The proach renders even more important efforts
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

ENCOURAGING WALKING
1139
3. Because a higher score indicates less agree-
ment, the expected signs of coefŽ cients in
the logit estimates will be the opposite of
those shown in Table 7.
by schools to have a presumption in favour
of children living nearby.
Three distinct factors were distilled from
the responses to the transport coping scale.
Alternative modelling strategies were ex-
plored which included using the attitudinal
statements directly as explanatory variables
and using the factor scores. In general terms,
the latter approach provided a more readily
interpretable basis for explaining mode
choice. To furnish our analysis of results, a
simulation exercise was conducted focusing
on the behavioural dimension. Tentatively, of
the three factors discerned, an emphasis on
general environmental awareness would
seem to be less useful for transport policy
initiatives than directing policy towards mod-
ifying attitudes related to car-centredness and
individual responsibility. Hence this study
suggests that sustainable transport research
efforts should be directed to evaluating
speciŽ c initiatives. In the context of the jour-
ney-to-school trip, a useful survey of relevant
trends and initiatives is provided in Dickson
(2000). These initiatives include various
forms of safer routes to school projects
around the UK that centre around supervised
group walking with organised rotas of
parental involvement. This seems to Ž t in
with tackling the core of the transport variant
of the NIMBY effect, which we might call
the OPCO effect. Such an effect is character-
ised by a general consensus for the need to
reduce private car trips—so long as it is
Other Peoples’ Cars Only.
4. An alternative approach would be to assume
that it was only when a viewpoint was to-
wards the extreme of possible responses that
it affected the choice of transport mode. To
examine this, all questions were also split
–
into a low (1 or 2), mid (3 5) and high (6 or
7) response and the low and high variables
were entered as dummies. Following the
same procedure as before, but independently,
these were whittled down to a core of
signiŽ cant variables. A good degree of com-
parability is shown between the two ap-
proaches—questions 27, 39, 51, 28, 32, 37,
42, 46, 50 and 56 were signiŽ cant in some
way (at the 5 per cent level) under either
approach.
References
ARMSTRONG, N. (1993) Independent mobility and
children’s physical development, in: M. HILL-
MAN (Ed.) Children, Transport and the Quality
–
of Life, pp. 35 43. London: Policy Studies In-
stitute.
ATASH, F. (1994) Redesigning suburbia for walk-
ing and transit: emerging concepts, Journal of
Urban Planning and Development, 120,
–
pp. 48 57.
BIRD, C. (1940) Social Psychology. New York:
Appleton-Century-Crofts.
BRADSHAW, R. (1995) Why do parents drive their
children to school?, TrafŽ c Engineering and
–
Control, 36, pp. 16 19.
BREZNITZ, S. (1986) Are there coping strategies?,
in: S. MCHUGH and M. T. VALLIS (Eds) Illness
Behaviour:
pp. 325 329. New York: Plenum Press.
A
Multidisciplinary
Model,
–
CERVERO, C. and RADISCH, C. (1996) Travel
choices in pedestrian versus automobile ori-
ented neighborhoods, Transport Policy, 3(3),
Notes
–
1. This would include details of the work loca-
pp. 127 141.
tion, working hours and incomes of all mem- DEPARTMENT OF THE ENVIRONMENT, TRANSPORT
bers of the household. Whilst this is an
ultimate aim, the current study concentrates
solely on the one short trip, on a partial
analysis basis.
AND THE REGIONS (1998) A New Deal for
Transport: Better for Everyone. Cm 3950. Lon-
don: TSO.
DEPARTMENT OF TRANSPORT (1995) Transport
Statistics Report: National Travel Survey 1992/
94. London: HMSO.
2. All three model speciŽ cations were investi-
gated. In estimating models of the form pre-
sented in Table 6, model (2) provided the DICKSON, M., (2000) Characteristics of the escort
best Ž t in terms of the number of accurate
predictions, reaching 80 per cent, compared
to 76 per cent from model (1) and 68 per
education journey, in: Transport Trends,
pp. 47 54. Department of the Environment,
Transport and the Regions.
–
cent for model (3) (the most difŽ cult on EDWARDS, A. L. (1957) Techniques of Attitude
these criteria, since there are 3 modes to
predict).
Scale Construction. New York: Appleton-Cen-
tury-Crofts.
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

COLIN BLACK ALAN COLLINS AND MARTIN SNELL
1140
,
EDWARDS, A. L. and KILPATRICK, F. P. A. (1948) LIKERT, R. (1967) The method of constructing an
attitude scale, in: M. FISHBEIN (Ed.) Attitude
A technique for the construction of attitude
–
Theory and Measurement, pp. 90 95. London:
scales, Journal of Applied Psychology, 32,
–
John Wiley.
pp. 374 384.
MADDISON, D., PEARCE, D., JOHANSSON, O. ET AL.
(1996) Blueprint 5: The True Costs of Road
Transport. London: Earthscan.
GOODWIN, P. (1992) A review of new demand
elasticities with special reference to short and
long run effects of price changes, Journal of
ORSKI, K. A. (1990) Can management of travel
demand help solve our growing trafŽ c conges-
tion and air pollution problems?, Transporta-
–
Transport Economics and Policy, 6, pp. 155
169.
GOODWIN, P. (1995) Summary: Car Dependence:
a report for the RAC Foundation for Motoring
and the Environment Royal Automobile Club.
London: RAC Foundation for Motoring and the
Environment.
–
tion Quarterly, 44, pp. 483 498.
OSGOOD, C. E., SUCI, G. J. and TANNENBAUM, P.
H. (1957) The Measurement of Meaning. Ur-
bana: University of Illinois Press.
OUM, T. H., WATERS, W. G. II and YONG, J. S.
(1992) Concepts of price elasticities of trans-
port demand and recent empirical estimates,
Journal of Transport Economics and Policy,
GUILFORD, J. P. (1954) Psychometric Methods.
New York: McGraw-Hill.
HALLMAN, W. K. and WANDERSMAN, A. (1992)
Attribution of responsibility and individual and
collective coping with environmental threats,
–
26, pp. 139 154.
PEIRSON, J., SKINNER, I. and VICKERMAN, R.
(1996) Estimating the external costs of U.K.
passenger transport: the Ž rst steps towards an
efŽ cient transport market, Environment and
–
Journal of Social Issues, 48(4), pp. 101 118.
HALLSWORTH, A. G. (1987) Some cross-national
differences in the perception of nuclear war
threat: The Netherlands, France, West Ger-
many, Tijdschrift voor Economische en Sociale
–
Planning A, 27, pp. 1977 1993.
PETTY, R. E. and CACIOPPO, J. T. (1986) Com-
munication and Persuasion: Central and Per-
ipheral Routes to Attitude Change. New York:
Springer-Verlag.
–
GeograŽ e, 78, pp. 290 296.
HENSHER, D. A., MCLEOD, P. B. and STANLEY, J.
K. (1975) Usefulness of attitudinal measures in
investigating the choice of travel mode, Inter-
national Journal of Transport Economics, 2,
POTTER, S. and HUGHES, P. (1990) Vital Travel
Statistics. London: Transport 2000.
–
pp. 51 75.
RAY, J. J. (1989) Acquiescence and problems with
forced-choice scales, Journal of Social Psy-
HILLMAN, M., ADAMS, J. and WHITELEGG, J.
(1990) One False Move … A Study Of Chil-
dren’s Independent Mobility. London: Policy
Studies Institute.
–
chology, 130, pp. 397 399.
RCEP (ROYAL COMMISSION ON ENVIRONMENTAL
POLLUTION) (1995) Eighteenth Report: Trans-
port and the Environment. Oxford: Oxford Uni-
versity Press.
HOUSTON, B. K. (1987) Stress and coping, in: C.
R. SNYDER and C. E. FORD (Eds) Coping with
Negative Like Events: Clinical and Social Psy-
RESER, J. P. (1980) Automobile addiction: real or
imagined?, Man-Environment Systems, 10,
–
chological Perspectives, pp. 373 399. London:
Plenum Press.
–
pp. 279 287.
HUGHES, P. (1993) Personal Transport and the
Greenhouse Effect. London: Earthscan.
JAMES, W. P. T. (1995) A public health approach
to the problem of obesity, International Journal
SLOVIC, P., FISCHHOFF, B. and LICHTENSTEIN, S.
(1982) Why study risk perception?, Risk Analy-
–
sis, 2(2), pp. 83 93.
STOKES, G. and HALLETT, S. (1992) The role of
–
of Obesity, 19(3), pp. 37 45.
advertising and the car, Transport Reviews, 12,
JOHNSON, M. A. (1978) Attribute importance in
–
pp. 171 183.
multiattribute transportation decisions, Trans-
THURSTONE, L. L. and CHAVE, E. J. (1929) The
Measurement of Attitude. Chicago, IL: Univer-
sity of Chicago Press.
–
portation Research Record, 673, pp. 15 21.
JONES, P., DIX, M., CLARKE, M. and HEGGIE, I.
(1983) Understanding Travel Behaviour.
Aldershot: Gower.
TOLLEY, R. S. (Ed.) (1997) The Greening of Ur-
ban Transport: Planning for Walking and Cy-
cling in Western Cities, 2nd edn. Chichester:
Wiley.
KHATTAK, A. J. and PALMA, A. DE (1997) The
impact of adverse weather conditions on the
propensity to change travel decisions: a survey
of Brussels commuters, Transportation Re-
ULLRICH, O. (1990) The pedestrian town as an
environmentally tolerable alternative to mo-
torised travel, in: R. S. TOLLEY (Ed.) The
Greening of Urban Transport: Planning for
–
search A, 31(3), pp. 181 203.
LEE-GOSSELIN, M. E. H. and PAS, E. I. (1997) The
implications of emerging contexts for travel
behaviour research, in: P. STOPHER and M.
LEE-GOSSELIN (Eds) Understanding Travel Be-
–
Walking and Cycling in Western Cities, pp. 97
109. London: Bellhaven Press.
WANG, K. A. (1932) Suggested criteria for writing
attitude statements, Journal of Social Psy-
–
haviour in an Era of Change, pp. 1 28. Oxford:
–
Pergamon.
chology, 3, pp. 367 373.
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014

ENCOURAGING WALKING
1141
WHITELEGG, J. (1993) Transport for a Sustainable
Future: The Case for Europe. London: Bel-
haven Press.
2020 vision, in: P. STONHAM (Ed.) Local Trans-
port Today and Tomorrow, pp. 9 16. London:
Local Transport Today.
–
WOOTON, J. (1993) Local transport solutions with
Downloaded from usj.sagepub.com at TRENT UNIV on October 16, 2014