Transport Policy 44 (2015) 9–18

Contents lists available at ScienceDirect

Transport Policy
journal homepage: www.elsevier.com/locate/tranpol

Children's travel to school—the interaction of individual, neighbour-

hood and school factors
Sue Easton n, Ed Ferrari
Department of Town and Regional Planning, University of Sheffield, Western Bank, Sheffield S10 2TN, UK

art ic l e i nf o a b s t r a c t

Article history: The increase in average distance from home to secondary school over recent decades has been accom-
Received 22 October 2014 panied by a significant growth in the proportion of pupils travelling to school by motorized means as
Received in revised form opposed to walking or cycling. More recently this switch in travel mode has received considerable at-
6 May 2015
tention as declining levels of physical activity, growing car dependence and the childhood obesity “crisis”
Accepted 30 May 2015
Available online 10 July 2015
have pushed concerns about the health of future generations up the public health agenda, particularly in
the U.S., but also in the UK and Europe. This has led to a proliferation of international studies researching
Keywords: a variety of individual, school and spatial characteristics associated with children's active travel to school
Travel to school which has been targeted by some governments as a potential silver bullet to reverse the trend. However,
Active transport
to date national pupil census data, which comprises annual data on all English pupils, including a mode of
Mode of travel
travel to school variable, has been under-utilised in the analysis of how pupils commute to school. Fur-
Multilevel model
Sociospatial thermore, methodologically, the grouped nature of the data with pupils clustered within both schools
Secondary and residential neighbourhoods has often been ignored - an omission which can have considerable
Pupils consequences for the statistical estimation of the model. The research presented here seeks to address
Motorised transport both of these points by analysing pupil census data on all 26,709 secondary pupils (aged 11–16) who
attended schools in Sheffield, UK during the 2009–10 school year. Individual pupil data is grouped within
school, and neighbourhood, within a cross-classified multilevel model of active versus motorised modes
of commuting to school. The results support the findings of other research that distance to school is key,
but suggest that sociospatial clustering within neighbourhoods and schools is also critical. A further
finding is that distance to school varies significantly by ethnicity, with white British pupils travelling the
shortest distance of all ethnic groups. The implications of these findings for education and transport
policy are discussed.
& 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).

1. Introduction 1975), which involves both longer travel distances and an urban
form that favours car use (Dieleman et al., 2002, Newman and
In the mid-1980s the mean distance travelled to school by 11– Kenworthy, 2006). A third factor that has also influenced the
16 year olds in the UK was just over 2 miles; by 2013 this had length of children's journey to school is legislation promoting
almost doubled, increasing to 3.7 miles (Department for Transport, parental choice, which has encouraged the selection of out-of-area
2013). This lengthening of the high school commute has been in- schools (see for example Parsons et al., 2000, Hoare, 1975). In re-
fluenced by some of the urban-structural processes which have cent studies it has been estimated that less than half of all school-
occurred over the past 50 years. Firstly a marked increase in the age children in England now attend their nearest school (Allen,
size of high schools, which began in the post-war decades (Rigby, 2007, Ferrari and Green, 2013).
1979) has resulted in secondary schools drawing their pupil in- These changes in the spatial configuration of schools and urban
takes from wider catchment areas on average. Second, the sub- space have been accompanied by significant social change such as
urbanisation and decentralisation which has occurred in many the rise of the dual-working family and growing private car
cities has dispersed some school-aged children to family housing ownership, a corollary of increased household affluence. These
in low density new-build housing estates on the outskirts (Hoare, have occurred over a period that has seen the cost of car travel
decrease in real terms compared to other forms of transport
n
Corresponding author. (especially following deregulation and privatisation of public
E-mail address: s.easton@sheffield.ac.uk (S. Easton). transport which occurred in the 1980s (Fairhurst and Edwards,

http://dx.doi.org/10.1016/j.tranpol.2015.05.023
0967-070X/& 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
10 S. Easton, E. Ferrari / Transport Policy 44 (2015) 9–18

1996)). The rise in volume of road traffic associated with increased travelling by car, bus, or train are totally “passive” or “non-active”
private car use has also led to rising concerns about road safety, (see for example Sirard and Slater, 2008, Voss and Sandercock,
which has in turn contributed to decreasing child independence 2010). However, this is not necessarily the case, particularly with
and increased parental surveillance. Parental strategies to cope regards to public transport where users walk to and from bus or
with this dual challenge often most conveniently involve driving tram stops or train stations (Rissel et al., 2012). Yet, whilst it is
children to school en route to work. acknowledged that the degree of activity involved in different
All of these factors have combined to produce a highly complex modes of transport can be conceptualised as a continuum, which
pattern of travel from home to school characterised by, and en- itself has significant policy implications, data considerations in the
abled by, growth in the use of motorised forms of transport. Ac- present study mean that we generally classify journeys into those
cording to 1975/6 National Transport Survey data for Great Britain, that are predominantly “active” or “motorised”.
55% of all secondary school pupils walked to school, and 7% tra- There is now a burgeoning international literature on active
velled by car (Rigby, 1979). By 2012 only 38% of pupils aged 11–16 commuting to school, particularly from the US, in the wake of a
years walked to school and 26% travelled by car (Department for childhood obesity ‘epidemic’, which has shone a spotlight on
Transport, 2013). In 1975–6, walking was the selected mode of school commuting as a potential ameliorative agent that could
travel for 93.6% of all “education” trips under 1.6 km (approxi- provide children with a regular daily dose of physical exercise
mately 1 mile), exemplifying the key underlying constraint on (Banerjee et al., 2014). Although there are considerable differences
modal choice: distance. between the case of the US and Europe in terms of local geo-
Notwithstanding the effect of distance, the choices that chil- graphy, school-siting, the level of car dependency and the pro-
dren (and their parents) make with regards to school commuting portions of children walking and cycling to school, the dramatic
may depend crucially on the interaction of several factors oper- decrease in active commuting witnessed across North America in
ating at a number of levels. Neighbourhood-level factors, which recent decades is one possible future scenario in the UK.
include characteristics of the urban form and structure, may have a A wide range of factors have been found to be associated with
range of direct and indirect effects on travel behaviour. School- active school commuting. Stewart’s (2011) review of 42 studies
level factors, most notably variations in the ‘performance’ of found 480 correlates including: distance to school, family income
schools and the socioeconomic composition of their pupil intake, (access to private transport), concern about traffic and crime en
may influence school and residential location choices, thereby route, parental views on walking, cycle use and family timetables.
potentially circumscribing travel options and average travel dis- Urban form has both a direct effect on mode of travel choice and,
tances to school. Individual-level characteristics, such as age, have by influencing parental opinion, an indirect effect. The urban form
a relationship to the extent to which children will countenance or factors Stewart (2011) identified from other studies include:
be empowered to choose active forms of travel. The relationship
between factors at these different levels is likely to be very com-
Active transport infrastructure-pavements, safe crossings, cycle
plex: individual pupils are simultaneously ‘members’ of their paths;
neighbourhood and the school they attend, and models of travel
Barriers such as major road or railway crossings encountered en
behaviour may be underpinned by both fixed (e.g., age, gender) route;
and random effects (e.g., distance to school).
Network connectivity-local streets, route choice, cul-de-sacs;
The aim of this paper is to specifically consider the interaction
Land use mix-residential (populated) versus industrial, parks,
of these effects in explaining the travel mode of choice for sec- derelict land;
ondary school children in Sheffield, UK. A typical UK city char-
Residential density – increased numbers of people, “eyes on the
acterised by a high degree of self-containment, significant social street”;
variation between schools and neighbourhoods, and a highly
“Walkability”-aesthetic environment (greenery, trees, etc).
heterogeneous set of pupils within the context of a ‘loosened’,
non-hierarchical spatial relationship between home and school The evidence of the impact of urban form is broadly mixed and
locations. The findings are important for policy makers aiming to is likely to be highly context specific. Kemperman and Timmer-
maximise the use of active forms of transport (e.g. for public mans (2014) found that Dutch primary school children were more
health reasons) or to minimise car use (e.g. for environmental or likely to walk (though not necessarily cycle) in more urbanised
congestion reasons) and suggest that policy efficacy is likely to be neighbourhoods, although the relationship between factors is
highly contingent on contextual factors, not only of individuals but complex and the impact of environmental characteristics may be
of the schools they attend and the neighbourhoods they live in. indirect (in that distance, for example, is a function of density).
Schlossberg et al. (2006) found that the density of road junctions
1.1. Structure of the paper and cul-de-sacs in a neighbourhood, as proxies of ‘walkability’,
were significant predictors of walking rates among middle school
The paper is organized as follows: Section 2 highlights some of pupils in Oregon. Urban form explanations can only be partial,
the shortcomings of the literature in this area to date. Data and however. Stead (2001) found that individual and household-level
Methods, are described in Section 3, and the Results of the mul- socioeconomic factors were more important than urban form in
tilevel models are presented and discussed in Section 4. The explaining travel patterns, although no attempt was made to
concluding remarks and policy implications are outlined in Section predict travel mode.
5. The literature is further complicated by the impact of age on
the results. It is widely understood that age is a significant cor-
relate of active commuting (see for example Aarts et al., 2013,
2. Active commuting: definition and correlates Johansson et al., 2012). However, previous studies comprise a wide
range of subject age-groups, which preclude systematic compar-
The majority of the literature on commuting to school focuses ison. There has tended to be a focus on younger children, who
on walking and cycling, which are generally referred to as “active” often have less independence (Mammen et al., 2012) and who live
modes of transport. This term is often used in an oppositional, closer to school on average. English primary school children aged
dichotomous sense which either explicitly states (see for example 5-10 live 1.6 miles from school on average, compared to 3.7 miles
Lee et al., 2008), or implies that modes of transport such as for 11–16 year olds (Department for Transport, 2013). In order to
 S. Easton, E. Ferrari / Transport Policy 44 (2015) 9–18 11

minimise the potential effect of parental supervision and varia- proportion of pupils who attend their nearest school2 was on
tions in the extent children may be allowed autonomy, this study average 48.8% (median ¼ 48.3%), but varied considerably
focuses on high school children aged 11–16. (min¼0.0%, max¼ 95.2%, s ¼27.3%). It is clear that children at-
Much of the literature focuses either on individual-level pre- tending schools that are not their nearest is very widespread, va-
dictors of commuting behaviour or on the effect of physical lidating the selection of a cross-classified model in which pupils
neighbourhood characteristics (urban form). Given that both are members simultaneously of schools and neighbourhoods.
strands of research appear to claim at least partial success in ex- A linked database comprising tables on pupil, schools and
plaining mode suggests that there are important interaction or neighbourhoods was created. Anonymised data on pupils was
multilevel effects that arise from pupils’ multiple membership of supplied by Sheffield City Council within a strict data-sharing
non-hierarchical groups (e.g. households, schools and neighbour- agreement. Pupil variables included: school attended at the time
hoods simultaneously). Although multilevel methods have been of the relevant pupil census3, age, gender, ethnicity, eligibility for
employed to investigate children’s mode of travel to school in the free school meals (FSM), a special educational need (SEN) flag for
Netherlands (Aarts et al., 2013; Bere et al., 2008), Australia (Trapp SEN-statemented or “School Action Plus” (as these two categories
et al., 2012) and Belgium (D’haese et al., 2011), to the best of the may affect the choice of school), together with the geographical
authors’ knowledge this is the first piece of research to use this coordinates of pupils’ residential location and attended school.
technique in the UK. This is important because there is a need for Importantly, the data included a variable recorded at a single time-
country-specific studies given substantial differences in education point on pupils’ mode of travel to school. This variable was man-
policy, urban policy, and levels and forms of urbanisation (Kem- dated in pupil census data by the Department for Education from
perman and Timmermans, 2014, Sirard and Slater, 2008). 2008–9 to 2010–11. The principle advantage of the variable is its
Aarts et al. (2013) found that low SES neighbourhoods were sub-population coverage which comprises the vast majority of
negatively correlated with active commuting but that high social pupils. Its disadvantages lie in having been recorded at a single
cohesion and perceived social safety (among other factors) were point in time with little by way of contextual information. Most
positively associated with walking and cycling. D’haese et al., significantly, the variable does not represent the variability of
(2011) used multilevel modelling to allow for clustering within home-school travel by direction of travel, potential variation
schools, but (probably due to small sample size) no school-level through the week, or seasonal change. Although one US study of
factors were entered into the model. Despite the inclusion of pupils aged 9–15 found that, for those children who lived less than
neighbourhood environmental factors such as aesthetic quality in a mile from school, the majority of pupils who commuted to
the model, no clustering by neighbourhood was undertaken. Trapp school by active means one day a week did so for all five days
et al. (2012) also adjusted for clustering within schools but not for (Martin et al., 2007), this cannot be verified in the present case.
pupils living in different residential neighbourhoods (which vary The data were reality-checked by local authority officers who ex-
‘substantially’ by socioeconomic status p.173). The issue of cross- pressed concern that cycling is probably under-represented be-
classification where pupils from different neighbourhoods attend cause the survey is undertaken in winter when daylight is minimal
the same school is not addressed. Although neighbourhood and weather conditions can be severe. Consequently, we do not
walkability factors such as street connectivity and traffic volume disaggregate walking and cycling in our results. It should also be
are in the model, they are attached to the school rather than area noted that data collected in a classroom situation may be affected
of residence. Bere et al. (2008) also took clustering between by peer-group dynamics or 'social desirability’ bias (Millward et al.,
schools but not neighbourhoods into account, despite finding
2013). Several research designs, such as those incorporating the
significant differences in school commuting patterns by ethnicity.
use of GPS tracking or travel diaries (DfT, 2013) can overcome
Therefore although the above studies have used components of
these limitations, but at the cost of sacrificing the scale of data
multilevel models they have not taken into account patterns of
collection. The use of survey data, such as the British Household
autocorrelation through sociospatial segregation of residential
Panel Study/Understanding Society allows more socioeconomic
areas into account.
context but the sample size would not be large enough to analyse
local neighbourhoods and the role of school-specific or urban form
factors. Consequently, it was concluded that despite several sig-
3. Data and methods
nificant shortcomings the value offered by capturing usual mode
of travel at the individual level offered significant analytical
The research design conceptualises pupils as having (multiple)
advantages.
membership of both schools and neighbourhoods thus they are
The pupil table comprised 26,709 pupils linked to 100 different
grouped within both. The relationship between schools and
‘neighbourhoods’ and attending one of 27 different state second-
neighbourhoods however, is complex, with pupils from one
ary schools in Sheffield. Following the exclusion of missing and
neighbourhood attending many schools, and pupils within a single
unclassifiable data, 25,798 cases remained (representing a loss of
school potentially hailing from many different neighbourhoods.
3.4% of the data). Even taking into account this data loss, the high
Pupils also possess personal socioeconomic characteristics (age,
level of study population coverage avoids problems such as sample
gender, and ethnicity). A cross-classified multilevel model is con-
bias endemic in many national sample surveys (see for example
structed to estimate the factors that best explain the variance in
Mcdonald, 2008). Furthermore the size of the study population
pupils’ ‘active’ and ‘motorised’ commuting to school using a binary
and inclusion of all pupils resident within an entire local authority
logistic function.
district enabled a multilevel analysis of all neighbourhoods and in
Analysis was performed on all pupils aged 11–16 attending a
this case, secondary schools across the city.
state-funded secondary school (including academies) in 2009-10
Data on the 27 schools came from the city council and the
in Sheffield, England, a relatively self-contained city dominated by
Department for Education’s EduBase2 web portal. Variables in-
state school provision1. In our dataset, the neighbourhood-level
cluded: the proportion of pupils in the school eligible for free
school meals (the FSM rate); the proportion of pupils recorded as
1
In 2009–10, 3.8% of children in Sheffield attended an independent school,
compared to an England average of 7.2% (Department for Education, 2010). Schools,
2
Pupils and their Characteristics. London., table 10a; 6.3% of schools in Sheffield As measured by walking network distance, not crow-fly distance.
3
were independent, compared to an England average of 9.7% (ibid., table 10b). Late January 2010.
12 S. Easton, E. Ferrari / Transport Policy 44 (2015) 9–18

having any special educational need (SEN rate); the proportion of Table 1
pupils from black and minority ethnic backgrounds (BME rate); Categories for the “usual mode of travel” variable 2009–10.
the proportion of pupils for whom English was an additional
‘Usual’ mode of N % Notes
language (EAL rate); mean school Key Stage 4 results (proportion travel
of eligible pupils achieving five GCSEs at grades A*-C, including
English and Maths); and the faith status of the school (Christian/ ‘Active’ modes 13,352 50.0
Walk 13,310 49.8 Excludes 265 cases (see below) whose
secular).
distance from home to school was 43
Neighbourhood characteristics were drawn from the 2011 miles
Census and other small area data. In order to avoid small cell sizes Cycle 42 0.2
within the multilevel model (see below), census output areas
(OAs)4 were aggregated into 100 neighbourhoods previously de- ‘Motorised’ modes 12,446 46.6
lineated by Sheffield City Council for policy analysis (Thomas et al., Bus–Public service 5392 20.2
2009). A bespoke weighted house-price index was also created Bus–School service 2386 8.9
Bus–Unknown type 1030 3.9 Likely to be a mixture of public, school
using Land Registry’s data on property sales (sales were pooled for
bus and minibus for pupils with SEN.
the period 2007–2011 and seasonally adjusted to overcome small Car 2644 9.9
cell counts). Carshare 441 1.7
For each pupil, their route to school for both active and mo- Taxi 93 0.3
torised modes was modelled using ArcGIS Network Analyst 10.1. Tram 454 1.7
Train 6 o 0.1
Route distances from home to school were calculated using the
Ordnance Survey Integrated Transport Network (ITN) for Sheffield,
Excluded from 911 3.4
including the new Urban Paths layer on the basis of shortest route.
analysis
These layers were downloaded from Digimap (Edina, 2014). As- Other-not known 643 2.4 Potentially mixed active/motorised
sumptions were made about overall walking speeds (2.5mph) and modes but primarily from 3 schools—
driving speeds on different types of roads at peak commuting unclassifiable.
times to determine the route. The calculation of walk-specific Missing data 3 o 0.1
Walking unlikely 265 1.0 Pupils whose journeys were 3þ miles
distances (using urban paths as well as roads) is an important from home—inaccurate home address?
innovation that overcomes significant limitations of previous re- Total 26,709 100.0
search by allowing model variables to more closely replicate actual
walking behaviour among pupils.
Several urban form variables which previous research had environment, and is also the point at which children of secondary
identified as being related to commuting mode of choice school age are allowed to claim a free bus pass from the local
(Schlossberg et al., 2006) were also created for Sheffield local authority. An hour’s walk has also been used as a reasonable cut-
authority district. These included: residential density (calculated off point for walking to school by other researchers (Mcdonald,
from Office for National Statistics postcode directory data), build- 2008).
ing density (from Ordnance Survey MasterMap), cul-de-sac den- Sheffield has a number of ethnic communities, many of which
sity, road junction density and network junction density (including comprise small numbers of people who are very unevenly spatially
junctions with urban paths)-all of which were calculated from distributed. Therefore for the purposes of analysis within the
Ordnance Survey Integrated Transport Network layers using multilevel model it was necessary to aggregate the detailed ethnic
techniques described in (Reference Suppressed for Review). categories provided in the pupil census data into four very broad
Table 1 shows the frequency of the usual mode of travel vari- categories: white British, white other (e.g. EU, Irish), non-white
able for Sheffield pupils aged 11–16. This variable was recoded into and “not known” in order to avoid the problem of common sup-
a dichotomous variable representing “active” (walking or cycling) port (see below).
versus “motorised” (bus, car, taxi, tram, train) modes of commut-
3.1. Correlation among school-level variables
ing to school for the purposes of analysis. Local advice from
council officers working with schools indicated that the method of
A significant degree of inter-correlation was found among the
administration for this question had varied across schools and
school-level variables, especially BME and EAL rates (0.97), FSM
time, resulting in inconsistencies in the quality of data recorded
and SEN rates (0.79), as well as the SEN and EAL rates (0.59) - all of
over the 3 year period it was collected. It was suggested that the
which were significant at p ¼ 0.01. Conversely, the school perfor-
optimal year for data quality and robustness was 2009–105.
mance results at Key Stage 4 (GCSE-level) were found to be in-
Therefore this was the year selected for cross-sectional analysis.
versely correlated with both SEN ( 0.74), FSM ( 0.79) and, to a
Several groups of cases were removed from the analysis. 643
lesser degree with EAL rates ( 0.17). A variable was therefore
cases in the ‘other’ category, which might refer to active or mo-
created to represent “cumulative” disadvantage at the school level
torised modes, were removed from the analysis. These cases pri-
by summing the FSM, SEN, and EAL rates (as BME and EAL so
marily related to pupils from three schools, suggesting that there
highly correlated). However this was not found to be significant
were localised problems in the administration of this question. To
and was dropped from the model at an early stage.
avoid potential bias these cases were excluded. A further 265 cases
Faith schools were found to have lower rates of pupils eligible
in the ‘walk’ category were found to involve estimated distances
for free school meals (FSM), a proxy for deprivation (t ¼4, d.f. ¼ 4,
greater than 3 miles, and therefore potentially indicative of an
p¼ 0.05). This may in part be due to the higher degree of control
inaccurate home address. These cases were also removed from the
some faith schools have over their pupil selection criteria (Allen,
analysis. The three mile cut-off was used as it would take ap- 2007).
proximately one hour at a sustained walking pace in an urban
3.2. Correlation among neighbourhood-level variables
4
Of which there were 1817 in Sheffield in 2011.
5
Personal communication with council officers working directly with schools Significant correlations were also found across a range census
collecting the data. data at the "neighbourhood" level. In particular, the proportion of
 S. Easton, E. Ferrari / Transport Policy 44 (2015) 9–18 13

people in an output area with no access to a car or van was found to 3.5. Model estimation and model fit
be highly correlated with other local indicators of socioeconomic
disadvantage such as: unemployment (0.91), the rate of people in Models were estimated using the Markov Chain Monte Carlo
basic (low skilled) occupations (0.89), household occupancy (MCMC) Bayesian technique (Leckie and Bell, 2013), with initial
(overcrowding) (0.80), and the proportion with no qualifications estimates calculated using iterative generalised least squares
(0.73). Given these correlations and previous research findings (IGLS) as starting points for the MCMC estimation. The Deviance
that lack of access to a car ‘… is the best single indicator of relative Information Criterion (DIC) diagnostic was used to assess para-
meter significance and the “goodness of fit” of the model (Browne,
deprivation’ (Voas and Williamson, 2001, p. 73), this indicator was
2005). This diagnostic criterion balances overall model fit with
selected as a key variable measuring socioeconomic disadvantage
parsimony by penalising for model complexity (Reference Sup-
for entry into the model.
pressed for Review). A general rule of thumb is that a reduction in
the DIC of more than 5 implies a variable is significant to the
3.3. Urban form variables overall model (CMM, 2007). All models were estimated using
MLwiN (Centre for Multilevel Modelling, 2014). The significance of
The importance of urban form variables is underscored by the the regression coefficients for fixed effects were compared and
distinctive geography of Sheffield. Although England’s fourth lar- formally tested using the Wald test in MLwiN, which produces a χ2
gest city, it is relatively self-contained with an over-bounded ad- statistic. Random parts of the model, however, such as the be-
ministrative geography. The city comprises a diversity of urban tween-neighbourhood and between-school variances were tested
by observing the reduction in the DIC (Leckie and Bell, 2013). A
archetypes ranging from dense inner-city terraces to very spacious
“null” model was estimated with individual pupils grouped within
suburban neighbourhoods and semi-rural villages. The city’s to-
the two higher level classifications-neighbourhoods and schools,
pography is hilly, which is likely to have an impact on travel mode
but with no explanatory variables (after Leckie and Bell (2013)).
choice. For these reasons, it was considered important to include a
Both of the higher-level variance components – (within school)
range of urban form variables in the model. The degree of corre- between-neighbourhood variance and (within neighbourhood) be-
lation found among the urban form variables (residential density, tween-school variance—were found to be highly significant (redu-
population density and building density) was also found to be cing the DIC by -5989 and -7831 respectively, thereby confirming
significant. Cul-de-sac density, and network junction density were the appropriateness of a cross-classified multilevel model.
also correlated. Small but significant correlations were also found
between distance to school and the majority of the urban form 3.6. Small cell sizes
variables, which were tested through the addition of interaction
terms in the model. One issue with complex multilevel analysis is that the number
of individuals (pupils in this case) is broken down into numerous
categories such as school, neighbourhood, age, gender, ethnic
3.4. A multilevel model of travel mode group, SEN status and so on which quickly results in small cell
sizes. Consequently variables such as detailed ethnic categories
A multilevel model was constructed in which the dependent and very small areas were aggregated up into larger groups and
variable was mode of commuting to school, dichotomised into neighbourhoods in order to ensure there were enough pupils in
active versus motorised means of transport. A multilevel model each category to undertake statistical estimation. However, the
was appropriate because of the grouped nature of the data small number of pupils in some categories still restricted further
(Duncan et al., 1996). The data violates the underlying principle of exploration of certain variables—such as the interaction between
standard ordinary least-squares regression models of in- pupils with an SEN flag and gender, detailed ethnic background
dependent, uncorrelated observations. In this case pupils are so- and gender.
cially and spatially grouped within both neighbourhoods and
schools. If grouping is ignored, this can result in an under-
estimation of the standard errors of regression coefficients, which 4. Results and discussion
means that statistical significance could be over-estimated (Centre
for Multilevel Modelling, 2007). In the cross-classified multilevel 4.1. Some descriptive statistics
model employed here pupils from the same neighbourhood can
The network distances between home and school ranged from
attend different schools, while pupils in the same school can come
less than 100 m to 21.7 km (13.5 miles), although the data are very
from many different neighbourhoods (Fig. 1).
skewed (see Fig. 2 below). The mean distance for all pupils was
2.7 km but the median home-school distance was just over 2 km.
The median home-school distance for pupils who walked to school
was 1.33 km (mean 1.38 km) and for those who used motorised
modes the median was 3.1 km (mean 3.9 km). For pupils living less
than one mile from school, 82% walked, although this represents a
very substantial decline over the last three decades when com-
pared to over 94% of high school pupils aged 11–17 in 1975/6
(Rigby, 1979). Despite this decline, walking rates in Sheffield are
slightly higher than the national average, with just under 50% of
high school pupils recorded as walking to school (Table 1), com-
pared to 38% across Great Britain (Department for Transport,
2010).
The results of the final model are presented in Table 2. The
Fig. 1. Schematic diagram of a cross-classified multilevel model showing non- most significant explanatory components in the model are be-
hierarchical membership of neighbourhoods and schools. tween-school variation, between-neighbourhood variation and
14 S. Easton, E. Ferrari / Transport Policy 44 (2015) 9–18

importance of taking the sociospatial clustering of individual pupils

into account in any examination of school travel patterns. Studies
such as those using survey sample data that do not group pupils
into neighbourhoods and schools may miss these higher-level ef-
fects and may have led to the over-estimation of the proportion of
variance explained by individual and urban form variables.

4.2. Route distance from home to school

After taking sociospatial clustering within schools and neigh-

bourhoods into account, distance from home to school was found
to be by far the strongest predictor of motorised travel, reducing
the DIC by 4184.6 This confirms findings from international studies
on travel to school in the U.S. and Sweden (Johansson et al., 2012,
Mcdonald, 2008) and earlier studies in Britain (Rigby, 1979). As
implied by the boxplot in Fig. 2 above, the mean distances for
active (1.5 km) and motorised (3.9 km) travel were significantly
different (t¼  98, df ¼ 25796, p ¼0.000, variances tested equal).
The mean distance from home to faith schools was significantly
Fig. 2. Boxplot of active versus motorised modes of travel to school. Note–Fig. 2. greater (3.45 km compared to 2 km for secular schools at
includes pupils who lived more than 3 miles away from school who were recorded p¼ 0.000), which is unsurprising given that the two Catholic High
as “walking”, most of whom were excluded from the data (see Table 1). Schools in Sheffield don’t have nominated de jure catchment areas
and therefore take pupils from across most of the city.
Table 2
Results of the final cross-classified multilevel model. 4.3. Gender

Variable Regression Standard error DIC Difference More boys walked or cycled to school than girls (the reference
coefficient in model fit
group). This finding accords with those studies reviewed by
(DIC)
Stewart (2011) where gender effects were identified. Although
girls travelled slightly further to school (2.72 km) than boys
Baseline DIC(in- 35,733 (2.67 km) on average, this difference was not found to be statis-
dividual pupil)
tically significant. However, further exploratory analysis showed a
Random Components
(Within school) 29,744  5989
significant difference between the proportion of girls and boys
Between neigh- who attended a faith school7 (χ2 ¼ 11.9 at 1 degree of freedom,
bourhood p¼ 0.001).
variance Excluding the two faith schools from the analysis, significantly
(Within 24,912  4832
fewer girls walked and significantly more girls travelled to school
neighbourhood)
Between school by car (including car-sharing). Previous research has highlighted
variance gender differences in the level of independence granted to chil-
Fixed variables dren by parents (Hillman and Adams, 1992) which may involve
Constant (intercept) þ2.09nnn 0.39 24,913 more restriction in public spaces and a higher level of supervision
Distance home–  2.32nnn 0.163 20,729  4184 (Stewart, 2011). Giving a child a lift rather than allowing them to
school (km) walk affords them less freedom to roam and provides increased
Gender:
Female (ref. cat) þ0.168nnn 0.035 20,712  17
opportunity for parental surveillance. Johansson et al. (2012) also
Male found that Swedish boys were significantly more likely to cycle
Age (11-16)  0.189nnn 0.026 20,696  16 than girls. Among the tiny number (N ¼ 42, 0.2%) of pupils who
Ethnic Group: (altogether) 20,548  148 cycled to school in the study population8 the ratio of boys to girls
Not known  0.29 0.243
was over 4:1.
White British (ref cat.)  0.286 0.15
White EU/Irish  0.727nnn 0.056
Non-white (all) 4.4. Age
Individual SEN Status  0.671nnn 0.147 20,530  18
[Statement or
Age was found to be positively correlated with distance tra-
School Action þ]
IA: Age  Distance to 0.066nnn 0.014 20,503  28
velled to school. Significant differences were found between the
School mean for 11 year olds (2.5 km) and older children-14 year olds
IA: Distance  0.301nnn 0.078 20,474  29 (2.74 km, p¼ 0.01) and 15-16 year olds (2.83 km, p ¼ 0.001). So
Cul-de-sac density

nnn
significant at p ¼0.001 6
Alternative models were also specified, including one in which distance to
school was allowed to vary. However, this and several other specifications could not
be computed because the existence of a negative definite V matrix, most probably
caused by the inclusion of continuous explanatory variables. These variables and
interactions between them were considered important hypothetically and were
distance from home to school. The improvement in model fit therefore retained; future work could consider experimentation with different
which resulted from the inclusion of these variables far outweighs categorical and non-linear explanatory variables.
7
Possibly due to the use of academic and other selection criteria among these
that of the other significant variables. This finding highlights the schools.
high levels of autocorrelation that exist across residential space as 8
Which may have been at its lowest in late January when the data was re-
well as within educational institutions, and the consequent corded due to short daylight hours and cold or inclement weather.
 S. Easton, E. Ferrari / Transport Policy 44 (2015) 9–18 15

pupils aged 14–16, on average, travelled further to school in 2009– Table 3

10. This is due to a cohort effect as the school choices available in Mean distance to school by broad ethnic category.
Sheffield when the children aged 14–16 in 2009–10 started high
Ethnic category N Mean distance from home to school
school (i.e. 2004–2006) were different from those available to (km)
pupils aged 11 in 2009–10 due to a planned school closure.
An interaction term age  distance to school was created in White British (Reference group) 20,890 2.5
“Not known” 142 2.75n
order to take account of this interaction. Once this interaction term
Non-white 5224 3.38nnn
was added into the model an inverse relationship between age and White Irish/EU 453 3.45nnn
walking to school was observed. This means that older high school
n
pupils were less likely, on average, to walk to school compared to nnn
Significant at the 0.05 level;
11–12 year olds once distance to school was taken into account. This Significant at 0.001 level

may be partly due to the well-documented increased changes in

sleep pattern associated with puberty (Carskadon, 2011, Foster and
Kreitzman, 2014) putting pressure on time in the mornings for Table 4
teenagers. Distance to school by detailed ethnic group.

Ethnic group–detailed categories N Distance to school (km)

4.5. Ethnic group
White British 20,890 2.50
The mean distance between home and school was tested for Not known 142 2.75
Other e.g. Yemeni 568 2.84
different broad ethnic categories using analysis of variance (AN- Mixed-White and Black 709 3.00
OVA) with a post-hoc Scheffe test (Table 3). All other ethnic groups Asian Pakistani 1459 3.05
for which data were available were found to have travelled sig- Mixed-White and Asian 526 3.10
nificantly further to school than white British pupils in Sheffield in Romany 27 3.24
White Irish/EU etc 453 3.45
2009–10.
Asian-other 717 3.64
Exploring variation in mean distances to school using the more Black-Afro-caribbean 243 3.97
detailed ethnic categories uncovered significant heterogeneity Black-African: Other 435 4.27
among pupils from Sheffield’s different ethnic communities (Ta- Black-African: Somali 440 4.35
ble 4). Of particular note is the fact that pupils from Black ethnic
groups travelled the furthest to school, at 4 km and over, com-
cycling to school. A t-test on mean distance to school showed that
pared to an average of 2.5 km for white British pupils. Further
these SEN pupils travelled significantly further than non-SEN pu-
analysis found that this was driven by a higher propensity for
pils (means of 3.9 km and 2.7 km respectively). As mentioned
pupils from the Afro-Caribbean and ‘African–Other’ groups to at-
above, children with these types of SEN may be required to travel
tend a Christian faith school, which has also been noted by other
further in order to access schools which are equipped to support
researchers (Weekes-bernard. 2007). Somali pupils, however, did
their particular needs. Although male pupils with this SEN flag
not attend Christian schools, probably due to Islam being their
outnumbered girls by 3:1, it was not possible to add an interaction
main religion of their community - yet on average these pupils
term between individual SEN status and distance to school or
travelled the furthest of all. This may be partly due to the clus-
gender into the multilevel model due to small pupil numbers.
tering of the Somali population in particularly deprived neigh-
bourhoods in inner-city locations9 with access fewer schools
4.8. Urban form variables
nearby.
Although there has been a significant focus on the impact of
4.6. Interaction between gender and ethnicity
urban form factors on commuting behaviour in recent years, the
model presented here suggests that urban form only explains a
A chi-squared test of mode of travel by gender by ethnic group
very limited amount of the variation between pupils who com-
indicated that fewer girls from certain ethnic groups were re-
mute by active and motorised means. The only urban form variable
corded as actively commuting to school. Among pupils of Asian
which was found to be significant was the density of cul-de-sacs
descent, significantly more girls travelled by car compared to boys
(no-through roads). Analysis showed that taking account of dis-
(χ2 ¼23, d.f. ¼ 11, p ¼0.05), although no significant difference was
tance to school, pupils who lived closer to school in areas of high
found in distance from home to school for Asian boys and girls. In
cul-de-sac density, were more likely to engage in active com-
some cultures girls may be perceived as more vulnerable or as muting. This finding is in opposition to findings from some studies
requiring greater ‘protection’ than boys. For example, Weekes-
in the US (see for example Schlossberg et al., 2006) and likely
bernard (2007) presents the concerns of several ethnic minority reflects the differences in the detailed layout of suburban arche-
parents specifically with regards to escorting their daughters to types. In the US dead-end roads are typically not through routes
school. for pedestrians. However, post-war suburban expansion in Shef-
field, in common with many English cities, was based on variants
4.7. Special educational need status of the “Radburn” principle (Womersley, 1954), where road cul-de-
sacs exist alongside a traversable network of walking routes.
Until now, special educational need does not seem to have been
included or described as a relevant explanatory factor in studies of 4.9. Variables not in the model
travel to school. However, the flag for SEN-statement or “school-
action plus” status, which applied to 457 pupils in this dataset, was The proportion of households without access to a car or van
found to be significantly inversely associated with walking or was found to be of borderline significance (it reduced the DIC by
only 4) with an inverse relationship to walking or cycling to
9
Following the arrival of many of their parents in Sheffield as refugees in need school. This implies that pupils living in areas with less access to
of social housing. private transport were in fact more likely to use motorised travel.
16 S. Easton, E. Ferrari / Transport Policy 44 (2015) 9–18

However, as outlined above, area levels of no/low car access have components.
been found to be correlated with core indicators of deprivation The core findings of this study are that mode of travel (active
(Voas and Williamson, 2001), and further exploration using ana- versus motorised commuting) is socially patterned within both
lysis of variance (ANOVA) exposed an interaction between ethni- schools and neighbourhoods. Rates of active commuting vary
city and levels of car ownership/access. Black Somali children were primarily by school, neighbourhood, and with the length of the
found to live in the most “deprived” neighbourhoods as indicated home-school commute. Furthermore distance to school has been
by the lowest levels of car access10. As noted above, Black Somali found to vary by pupil age and ethnic background. Motorised/ac-
pupils travelled the furthest to school on average. Pupils from the tive transport has been shown to vary by age, gender, ethnicity and
lowest car access neighbourhoods were significantly less likely to parental views on child independence and the need for escort
cycle or walk, take a dedicated school-bus (many of which serve (which may be related to perceptions of neighbourhood safety).
faith schools), to car-share or take a tram, and more likely to more In order to be successful, it is likely that transport policy goals
likely to travel by car, use a public bus or bus of “unknown type”, need to be grounded in the everyday reality of pupils’ travel to
or travel by “other” means. school which includes taking account of route distance to school,
Pupil eligibility for free school meals (FSM status) was not family schedules, parental concerns about road safety, neigh-
found to be significant and school faith status and mean school bourhood safety and child safety. And importantly, given that
GCSE performance (Key Stage 4 statistics) were not found to be transport choices vary by neighbourhood, pupil gender, age and
significant over and above the between school variation in the ethnicity, blanket policies across entire schools or cities may not
model at the higher level. be appropriate. In their stead more flexible and responsive mea-
sures need to be developed that aim to meet the expressed need of
individual communities. For example, the possibility of organising
5. Discussion and policy implications appropriately chaperoned walking school buses specifically for
girls might be explored jointly with some ethnic communities,
The findings in this paper suggest that the predictors of active where education escort is gender-biased. At the very least, it
school commuting are complex and conditioned by neighbour- suggests that detailed transport planning associated with schools
hood- and school-level factors as well as individual or household needs to recognise these factors. Whilst education planning in
characteristics. Initiatives promoting active commuting through, England does require an element of cooperation with transport
for example, transport and urban design improvements will have planners at the individual school level (e.g. through the production
only limited effect if societal and sociospatial structures oblige of statutory school travel plans), it is likely that better links be-
some groups of pupils to commute long distances-yet this is the tween education, health, transport and land use policies (for
corollary of policies promoting school choice. The mismatch be- housing planning) are needed at the ‘macro’ (e.g., city-wide) level.
tween education and transport policies has been discussed with While the government’s Travelling to School Initiative had some
reference to the air quality impacts of school choice by Marshall success, its evaluation found that improvements to child health
et al. (2010); this study provides further evidence of the mismatch were far from transformative—one barrier, according to local sta-
that can arise from school choice programmes. keholders, being that travel distances implied by increased school
It is also important to acknowledge that the home-school choice (DfE and DfT, 2010). Notably, revised statutory guidance for
commute is at the juncture of a number of different policy areas. school travel planning in England has resulted in a less pre-
Joint working across education, housing and health policy do- scriptive framework aimed at better reflecting ‘the requirements
mains as well as transport will be essential to make any real im- of today’s dynamic and increasingly autonomous schools system’
pact in terms of modal shift to active commuting, transport sus- (DfE, 2014, para 2.1).
tainability and associated future budgets for transport subsidies. In the context of the considerable distances that some children
For example, in terms of distance to school, optimal school-siting travel to school, walking all the way to school is unlikely to be
within appropriate population centres, school size, residential realistic aim (Mcdonald, 2008). Mixed-mode objectives may be
planning and development and policies on parental choice are all more appropriate in such circumstances: for example, campaigns
key factors. Policies on selective schools such as grammar and that encourage bus commuters to “walk an extra stop”, possibly in
denominational schools are also implicated, as these often draw combination with behavioural change incentives linked to bus fare
students from much wider catchment areas (Rigby, 1979, Taylor, structures. On the other hand, educational travel subsidies may
2002, Parsons et al., 2000). provide perverse incentives, such as families deliberately opting
Currently many educational policies are working in opposition for more distant schools in order to qualify for a subsidised bus
to sustainable transport goals of local travel and low carbon cities pass. More in-depth, qualitative research into the reasons that
by driving system-wide patterns of ‘excess commuting’ (Horner, children travel long distances to school would be helpful to shed
2002) to more distant schools. Such policies may also run counter light here.
to public health objectives to increase physical exercise among Given the importance of socioeconomic factors associated with
children and reduce exposure to environmental pollution. Policies distance travelled to school and modal choice, it is likely that
aimed at promoting active travel that do not recognise the inter- transport policies will impact differently on the educational out-
mediate benefits of modal switch to public transport (e.g. buses comes of diverse sub-populations. Transport is not simply reacting
and trains) may be unrealistic in a context of further school de- to ‘demand’ but rather an integral to societal frameworks of op-
centralisation, consolidation, and parental choice. A key priority, portunity and constraint within which education choices are ar-
therefore, is to ensure that transport planners who aim to en- ticulated. Increased pressure on public funding has already led to
courage modal shift from private to public transport do not miss the withdrawal of discretionary transport funding such as free bus
the opportunity to work in tandem with public health officials to passes for children attending faith schools in some areas (Sheffield
focus on the benefits of public transport as a ‘mixed-mode’ form of City Council, 2014). Since such schools may provide a viable al-
commuting which comprises both active and motorised ternative route to higher-performing schools for low-income fa-
milies (Ferrari and Green, 2013), it seems likely that such decisions
10
Probably due to the historical placement Somali refugees into social housing would disproportionately affect those families and contribute to
in central areas of the city. worsening educational inequalities.
 S. Easton, E. Ferrari / Transport Policy 44 (2015) 9–18 17

6. Conclusion Centre for Multilevel Modelling, 2007. Introduction to multilevel modelling

workshop. CMM. University of Bristol, Bristol.
Centre for Multilevel Modelling, 2014. MLwiN: A software package for fitting
This paper has demonstrated that a range of individual pupil, multilevel models. CMM. University of Bristol, Bristol.
school-level and neighbourhood level factors are associated with D’haese, S., de Meester, F., de bourdeaudhuij, I., Deforche, B., Cardon, G., 2011.
walking or being driven to school. Consistent with other studies, Criterion distances and environmental correlates of active commuting to school
in children. Int. J. Behav. Nutr. Phys. Act., 8.
the distance between home and school is found to be the most Department for Education, 2010. Schools, Pupils and their Characteristics. London.
significant individual-level factor by far, although there are im- DfE, 2014. New Home to School Travel and Transport Guidance (consultation
portant correlates of distance that suggest systematic socio- published 25 March 2014). Department for Education, London.
Department for Transport, 2010. National Travel Survey: England 2010. Statistical
economic variation in the geography of school commuting. At a Release-National Travel Survey.. Department for Transport, London.
time when less than half of all English pupils attend their nearest DfT, 2013. National Travel Survey: England 2013. Statistical Release Department for
school, rates of active commuting are likely to decline further. Transport, London.
DfE and Dft, 2010. An Evaluation of the ‘Travelling to School Initiative’ Programme—
Given the public health benefits of increased active commuting
Final Report. Department for Education, Department for Transport,
among children and reduced “excess” commuting by motorised Birmingham.
means, serious consideration is needed of the implications and Dieleman, F.M., Dijst, M., Burghouwt, G., 2002. Urban form and travel behaviour:
micro-level household attributes and residential context. Urban Stud. 39,
contradictions between education policies that promote parental
507–527.
choice, public health policies seeking to reduce childhood obesity/ Duncan, C., Jones, K., Moon, G., 1996. Health-related behaviour in context: a mul-
increase levels of physical activity, and environmental policies tilevel modelling approach. Soc. Sci. Med. 42, 817–830.
aimed at reducing pollution and promoting sustainable transport. Edina, 2014. Digimap Resource Centre. University of Edinburgh, Edinburgh.
Fairhurst, M. and Edwards, D. 1996. Bus Travel Trends in the U.K.
Ferrari, E.T., Green, M., 2013. Travel to school and housing markets: a case study of
6.1. Limitations and future research Sheffield, England. Environ. Plan. A 45, 2771–2788.
Foster, R.G., Kreitzman, L., 2014. The rhythms of life: what your body clock means to
you!. Exp. Physiol. 99, 599–606.
The key limitation to studies of the type reported in this paper Hillman, M., Adams, J., 1992. Children’s Freedom and Safety. Child. Environ. 9 (2),
is the lack of hard, measured data on the actual journeys to school 12–33.
made by pupils. As noted at the outset, the model of travel variable Hoare, R.A., 1975. The Location Efficiency of Secondary Schools. Institute of British
Geographers, Transport Group.
employed in this study has several limitations, although at present Horner, M.W., 2002. Extensions to the concept of excess commuting. Environ. Plan.
it remains the most appropriate means by which travel mode can A 34 (3), 543–566.
be analysed within individual level models. It is important to note Johansson, K., Laflamme, L., Hasselberg, M., 2012. Active commuting to and from
school among Swedish children-a national and regional study. Eur. J. Public
that the question was removed from the School Census after 2011, Health 22, 209–214.
limiting the potential for future work in this area. Any future Kemperman, A., Timmermans, H., 2014. Environmental correlates of active travel
studies aiming to consider the determinants of school travel be- behavior of children. Environ. Behav. 46, 583–608.
Leckie, G., Bell, A., 2013. Module 12: Cross-Classified Multilevel Models. Centre for
haviour aimed at evaluating the impact of competing national Multilevel Modelling, Bristol.
policies would benefit from the systematic inclusion of a small set Lee, M., Orenstein, M., Richardson, M.J., 2008. Systematic review of active com-
of pupil-level school travel indicators within the pupil census data. muting to school and children’s physical activity and weight. J. Phys. Act.
Health, 930–949.
There remains scope to further develop the specification of multi-
Mammen, G., Faulkner, G., Buliung, R., Lay, J., 2012. Understanding the drive to
level models to ensure that they are robust. escort: a cross-sectional analysis examining parental attitudes towards chil-
dren’s school travel and independent mobility. BMC Public Health, 12.
Marshall, J.D., Wilson, R.D., Meyer, K.L., Rajangam, S.N., Mcdonald, N.C., Wilson, E.J.,
2010. Vehicle emissions during children’s school commuting: impacts of edu-
Acknowledgements cation policy. Environ. Sci. Technol. 44 (5), 1537–1543.
Martin, S.L., Lee, S.M., Lowry, R., 2007. National prevalence and correlates of
walking and bicycling to school. Am. J. Prev. Med. 33 (2), 98–105.
The authors would like to acknowledge the support of the ESRC
Mcdonald, N.C., 2008. Children’s mode choice for the school trip: the role of dis-
(Grant ref.: ES/K003720/1) in funding the research on which this tance and school location in walking to school. Transportation 35, 23–35.
article is based through the Secondary Data Analysis Initiative. Millward, H., Spinney, J., Scott, D., 2013. Active-transport walking behavior: desti-
Special thanks are due to staff at Sheffield City Council, in parti- nations, durations, distances. J. Transp. Geogr. 28 (0), 101–110.
Newman, P., Kenworthy, J., 2006. Urban design to reduce automobile dependence.
cular Mark Pearce, Martin Fox and Jeremy Hamm, for their assis- Opolis. Int. J. Suburban Metropolitan Stud. 2, 35–52.
tance in the progress of many aspects of the research project. Parsons, E., Chalkley, B., Jones, A., 2000. School catchments and pupil movements: a
Eddie Bulpitt of Ordnance Survey provided invaluable help with case study in parental choice. Educ. Stud. 26, 33–48.
Rigby, J.P., 1979. Review of research on school travel patterns and problems. TRRL
the integration of the urban paths data to the network model. Supplementary Report. Transport and Road Research Laboratory, Great Britain.
Responsibility for the research, its interpretation and any errors or Rissel, C., Curac, N., Greenaway, M., Bauman, A., 2012. Physical activity associated
omissions naturally rests with the authors. with public transport use: a review and modelling of potential benefits. Int. J.
Environ. Res. Public Health 9, 2454–2478.
Schlossberg, M., Greene, J., Phillips, P.P., Johnson, B., Parker, B., 2006. School trips:
effects of urban form and distance on travel mode. J. Am. Plan. Assoc. 72,
References 337–346.
Sheffield City Council, 2014. Faith School Transport Consultation. Sheffield City
Council, Sheffield Available: 〈https://www.sheffield.gov.uk/education/informa
Aarts, M.-J., Mathijssen, J.P., Van OERS, J.M., Schuit, A., 2013. Associations between tion-for-parentscarers/schooltravel/faith-school-transport-consultation.html〉
environmental characteristics and active commuting to school among children: (Accessed 01.09.2014).
a cross-sectional study. Int. J. Behav. Med. 20, 538–555. Sirard, J.R., Slater, M.E., 2008. Walking and bicycling to school. Rev. Am. J. Lifestyle
Allen, R., 2007. Allocating pupils to their nearest secondary school: the con- Med. 2, 372–396.
sequences for social and ability stratification. Urban Stud. (Routledge) 44, Stead, D., 2001. Relationships between land use, socioeconomic factors, and travel
751–770. patterns in Britain. Environ. Plan. B-Plan. Des. 28, 499–528.
Banerjee, T., Uhm, J., Bahl, D., 2014. Walking to school: the experience of children in Stewart, O., 2011. Findings from research on active transportation to school and
inner city Los Angeles and implications for policy. J. Plan. Educ. Res. 34 (2), implications for safe routes to school programs. J. Plan. Lit. 26 (2), 127–150.
123–140. Taylor, C., 2002. Geography of the “New” Education Market: Secondary School
Bere, E., van Der horrst, K., Oenema, A., Prins, R., Brug, J., 2008. Socio-demographic Choice in England and Wales. Ashgate, Aldershot, UK.
factors as correlates of active commuting to school in Rotterdam, the Nether- Thomas, B., Pritchard, J., Ballas, D., Vickers, D., Dorling, D., 2009. A tale of two cities:
lands. Prev. Med. 47, 412–416. the Sheffield project. Social and Spatial Inequalities Research Group. University
Browne, W.J., 2005. MCMC Estimation in MLwiN (Version 2.0). Centre for Multilevel of Sheffield, Sheffield.
Modelling, Bristol. Trapp, G.S.A., Giles-Corti, B., Christian, H.E., Bulsara, M., Timperio, A.F., Mccormack,
Carskadon, M.A., 2011. Sleep in adolescents: the perfect storm. Pediatr. Clin. North G.R., Villaneuva, K.P., 2012. Increasing children’s physical activity: individual,
Am. 58, 637–647. social, and environmental factors associated with walking to and from school.
18 S. Easton, E. Ferrari / Transport Policy 44 (2015) 9–18

Health Educ. Behav. 39, 172–182. Decision-making among Black and Minority Ethnic Parents. The Runnymede
Voas, D., Williamson, P., 2001. The diversity of diversity: a critique of geodemo- Trust, London.
graphic classification. Area 33, 63–76. Womersley, J.L., 1954. Some housing experiments on radburn principles. Town
Voss, C., Sandercock, G., 2010. Aerobic fitness and mode of travel to school in Plan. Rev. 25, 182–194.
English schoolchildren. Med. Sci. Sports Exerc. 42, 281–287.
Weekes-bernard, D., 2007. School Choice and Ethnic Segregation—Educational