Content uploaded by Glen Bramley
Author content
All content in this area was uploaded by Glen Bramley on Jun 12, 2014
Content may be subject to copyright.
1 Introduction
There has been much debate about how to shape urban development in order to
achieve greater sustainability. Many claims have been made for the greater `sustain-
ability'of compact cities, although these have been particularly controversial in relation
to the social functioning and acceptability of urban communities (Jenks et al, 1996,
page 11). Urban forms cannot be considered sustainable if they are not acceptable to
people as places in which to live, work, and interact, or if their communities are
unstable and dysfunctional. Recognition of this has become a significant feature of
recent British urban and planning policy (HM Government, 2005; ODPM, 2003).
However, the `compact city versus urban sprawl' debate of the 1990s and early 2000s
has relied a good deal on assertion and assumption with regard to the social impacts of
different urban forms, with limited reference to systematic contemporary empirical
evidence (Jenks et al, 1996).
The main aim of this paper is to provide a comprehensive analysis of the impact
of differing urban forms on social sustainability, in relation to medium-sized existing
Social sustainability and urban form: evidence from five
British cities
Glen Bramley
School of the Built Environment, Heriot-Watt University, Edinburgh EH14 4AS, Scotland;
e-mail: g.bramley@sbe.hw.ac.uk
Nicola Dempsey
Oxford Institute for Sustainable Development, Oxford Brookes University, Oxford OX3 0BP,
England; e-mail: ndempsey@brookes.ac.uk
Sinead Power
Communities Analytical Services, Scottish Government, Edinburgh, Scotland;
e-mail: sinead.power@scotland.gsi.gov.uk
Caroline Brown, David Watkins
School of the Built Environment, Heriot-Watt University, Edinburgh EH14 4AS, Scotland;
e-mail: c.j.brown@hw.ac.uk, d.s.watkins@sbe.hw.ac.uk
Received 10 March 2008; in revised form 3 September 2008; published online 18 June 2009
Environment and Planning A 2009, volume 41, pages 2125^ 2142
Abstract. Planning and urban policies emphasise `sustainability', but claims that `compact cities' are
more socially sustainable and acceptable have been controversial and subject to limited empirical
testing. After a brief review of the concepts and debate, we set out new empirical evidence based on
household surveys linked to neighbourhood physical, map-based, and sociodemographic data for five
British cities. Statistical models are developed to account for systematic variations in the main social
sustainability outcomes. The results are considered both in terms of the role of particular urban form
and locational measures, but also in terms of the broader patterns of effects of packages of measures.
Outcomes relating to residential satisfaction, stability, neighbourhood environment, and safety are all
shown to be lower in higher density/central places, but it is also shown that a good deal of this
apparent effect is due to social and demographic factors. Interaction with neighbours and participa-
tion in groups is better at medium densities, controlling for other factors, while use of local services
is, as expected, greater in denser, more central locations. These findings indicate that compact cities
are not `win ^ win' on all dimensions of sustainability but, rather, that reductions in transport
emissions will have to be weighed against social criteria. In addition, urban form has different aspects,
which have differing social effects, and this knowledge could inform the future design of `smarter'
urban environments.
doi:10.1068/a4184
British cities, drawing on recent research.The specific objectives are (a) to translate the
concepts of social sustainability into operational outcome measures; (b) to measure
their incidence for a representative set of areas in our sample cities; (c) to relate these
patterns to neighbourhood and wider scale urban form measures, while allowing for the
influence of demographic and socioeconomic factors; and (d) to relate these findings
to the wider literature on urban form and sustainability, drawing out pointers both for
planning policy and for future research.
The paper begins (in section 2) with a brief review of the existing literature and
debates about urban form and social sustainability, highlighting some expected rela-
tionships and some areas of dispute and uncertainty. Section 3 describes the way in
which we carried out the research: linking a household survey to neighbourhood
physical, map-based, and sociodemographic data; and developing regression models
to account for systematic variations in the main social sustainability outcomes. The
results are presented in section 4. This highlights both the role of particular urban form
and locational measures and the broader patterns of effects of packages of measures.
Section 5 discusses the findings and some of their possible implications for policy.
Some limitations to the work should be noted at this point. The scope of the
empirical research reported here is limited to five medium-sized existing British cities,
suggesting that some care might be needed in extrapolating conclusions to conditions
found, at one extreme, in Inner London, or at the other extreme in rural and village
settlements. However, in a companion piece Bramley and Power (2009) use national
secondary data to analyse some of the same relationships across the whole of England,
including these extremes. In this paper we do not attempt to enlarge, either, on the
conceptual debate around the definition and significance of `social sustainability',
although again this has been explored in related work (Dempsey et al, forthcoming).
We take the range of interpretations derived there as given, merely noting their impor-
tance in contemporary policy rhetoric, and asserting that they provide a reasonable
representation of the range of social issues raised in the literature.
2 Existing literature on social sustainability and urban form
2.1 Social sustainability
Following reviews of the issue, based on both academic and policy literature (Bramley
and Power, 2009; Bramley et al, 2006; Dempsey et al, 2009), we propose that `social
sustainability' should be seen as comprising two main dimensions.
.Social equity, with particular reference (in the context of urban form) to access
to services and opportunities
essential local services such as shops, schools, health centres;
recreational opportunities, open space;
public transport;
job opportunities;
affordable housing.
.sustainability of community, comprising a number of subdimensions including:
pride in and attachment to neighbourhood;
social interaction within the neighbourhood;
safety/security (vs risk of crime, antisocial behaviour);
perceived quality of local environment;
satisfaction with the home;
stability (vs residential turnover);
participation in collective group/civic activities.
It is clear that, while each of these may be regarded as conceptually distinct, there will
often be reinforcing relationships between them. For example, where neighbourhoods
2126 G Bramley, N Dempsey, S Power, C Brown, D Watkins
are safe people may be more likely to interact with neighbours and this may lead to
more participation in collective activities.
While we consider these aspects of people's experiences under the umbrella term
`social sustainability', it must be recognised that other umbrella terms are used in
relation to some of the same or overlapping phenomena. For example,`social exclusion'
may entail lack of access to services, among other things (Pantazis et al, 2006; Pierson,
2002); the concept of `social capital' generally entails social interaction, trust (under-
pinned by safety), and civic participation, and is widely argued to be encouraged
by stability (Forrest and Kearns, 2001; Putnam, 1993); `social cohesion' also entails
social interaction, civic/collective participation, `social order' (again related to safety),
and place attachment (Forrest and Kearns, 1999). Some writers and official documents
refer to the grouping of factors we have labelled `sustainability of community' as
`quality of life'.
Some commentators would regard health and well-being as an important compo-
nent of social sustainability or the quality of life. We would agree that health and
well-being are important, but believe that many of the ways in which urban form
affects these outcomes will be through the factors which we have identified above.
Social interaction, safety, local environmental quality, and access to services are likely
to affect well-being directly as well as indirectly: for example, through opportunities for
exercise in daily life. Negative outcomes on these factors are likely to be implicated in a
higher incidence of mental ill-health. However, this study was not designed to provide
direct or detailed measures of health and well-being.
In this paper we examine measures of all of the above subdimensions of sustain-
ability of community, while concentrating on summary measures of access to/usage of
local services. Both access to jobs and access to affordable housing raise issues about
wider `market areas' within which people may search for and obtain such resources
and, as such, require a somewhat different scale of analysis from the essentially
neighbourhood focus of this article. It is broader scale aspects of urban structure which
are likely to be more important here as, for example, in the well-known `mismatch' thesis
about urban labour markets (Arnott, 1998; Houston, 2005; Ihlanfeldt and Sjoquist, 1998).
Some measures of public transport quality feature in the analysis here, but a fuller analysis
of transport networks and accessibility again goes beyond the scope of this paper (for
parallel work on this, see Ferguson and Woods, 2009).
2.2 The effects of urban form
It is clear from the literature that there are competing claims about the extent to which
urban form influences social sustainability: claims and debates which have, to date,
rarely been supported by empirical evidence (Jenks et al, 1996). Of the elements of
urban form which might be considered, density is the one that has received the most
attention with regard to its social impact. Much of this focus has been upon the policy
question of whether we should contain urban development by developing at higher
densities, or allow urban expansion and building at lower densities
ö
the `compact city'
versus `sprawl' debate (Barton, 2000; Breheny, 1992a; 1992b; DETR, 1999a; Ewing,
1997) and the related `new urbanism' literature (Calthorpe, 1993; CNU, 2004; Katz,
1994).
The density of urban development has the potential to impact upon all of the
dimensions of social sustainability. For example, higher densities may make access to
services and facilities both easier and more economically viable (Bunker, 1985; Burton,
2000a; 2000b; Collie, 1990; Haughton and Hunter, 1994; ODPM, 2003; Williams, 2000).
Higher densities may also mean that people are more likely to meet each other on the
street than they are in lower density areas (Duany and Plater-Zyberk, 2001; Talen,1999).
Social sustainability and urban form: evidence from five British cities 2127
In contrast, lower densities reduce the potential for spontaneous interaction and lead to
an orientation towards car travel (TCRP, 1998). Glynn (1981) and Nasar and Julian
(1995) found `sense of community' to be higher in neighbourhoods which facilitated
face-to-face interaction. There are, however, alternative arguments that in higher
density settings people may withdraw from social contact. Wirth (1938) argued that
high-density living, along with the anonymity of city life, leads to an increase in stress
and the severing of traditional ties resulting in a decline in community or social ties.
Bridge (2002, page 4) refers to Simmel's (1995) discussion of the `psychic overstimula-
tion of the city'. Hence, higher densities may lead to weaker social ties, at least beyond a
certain level (Freeman, 2001).
It is argued that in a compact city with high-density and mixed uses, communities
are likely to be more mixed and, as a result, have a lower level of social segregation.
Suburban sprawl in particular has come to be associated with high levels of segrega-
tion and inner-city decay (Bramley and Morgan, 2003; Burton, 2000a; CEC, 1990).
However, it is not axiomatic that social mix correlates with density or use mix, and
other factors such as housing type and tenure may be more significant than density
(Burton, 2000a).
As well as affecting social interaction, density may affect the appearance and
aesthetics of places and hence people's attachment to and pride in their neighbour-
hood. The Transit Cooperative Research Program (TCRP) review (1998) found little
evidence to suggest that Americans find sprawl less attractive than more compact
forms of development, although they do cite work by Nelessen (1994), Shore (1995),
and Diamond and Noonan (1996) which argues that lower density development is less
aesthetically pleasing [Audirac and Zifou (1989), on the other hand, argue the opposite].
Gordon and Richardson (1997) claim that, given the choice, people prefer low-density
suburban living to high-density urban living, noting that many consumer-preference
surveys have shown such a preference (findings which are echoed, in some senses, in our
own empirical findings). However, it is appropriate to `unpack' these suburban and dense/
urban living scenarios and identify which elements contribute to this general pattern,
whilst recognising that some individuals and groups may have different preferences.
Feelings of safety (from crime or antisocial behaviour) arguably enhance trust and
reciprocity between residents and contribute to the sense of community and sense of
place in a neighbourhood (Nash and Christie, 2003, page 15; Shaftoe, 2000, page 231).
Some of the claimed associations between safety and the built environment include the
cited benefits of natural surveillance, for example, windows directly overlooking streets
(after Alexander et al, 1977; Jacobs, 1961), which is a function of detailed urban design
rather than of density per se. Poor condition and maintenance of the built environment
are also claimed to have detrimental psychological effects on people's sense of safety
(DETR, 1999b; Woolley, 2002; Worpole, 2003). The idea of ``nobody car[ing]'' is closely
linked to the concept of the ``broken window syndrome'', where even ``cosmetic damage
can invite more serious anti-social or even criminal behaviour'' (Wilson, 1985, page 78;
Nash and Christie, 2003, page 47). Current government policy emphasises the impor-
tance of considering crime prevention as part of the urban design process (ACPO,
undated; ODPM/Home Office, 2004).
This brief review of the literature reveals that the relationship between density and
social sustainability appears to be rather complex, with at times contradictory hypoth-
eses or findings. There are reasons to expect access to services to be better in denser
urban forms, while quality of neighbourhood environment, community, and safety may
be less good in denser areas. The relationships for social interaction or participation
are less clear-cut in the literature. Interactions between urban form and social com-
position factors, including those associated with housing tenure, may be significant.
2128 G Bramley, N Dempsey, S Power, C Brown, D Watkins
For example, certain inner city higher density areas may specialise in housing groups
such as students or young professionals who display high mobility and fewer neighbour-
hood-based social networks (Bailey and Livingstone, 2007; Bramley and Morgan, 2003).
One of the problems with trying to relate urban form to social phenomena is the
difficulty of separating causal effects from selection effects, where selection effects are
the result of different individuals and groups living in different places. For example,
people with mobile careers or lifestyles may tend to live in certain areas because of their
housing tenure, access to central amenities, and their affinity to similar people. They
may display low place attachment, community engagement, or local social interaction,
and high mobility, but this is not causally related to housing types or neighbourhood
density. It is much more likely to be related to their individual characteristics, some of
which may be measurable (age, occupation, household type) but some of which may not.
It is difficult to separate these factors fully in cross-sectional analysis, although we can
go some way by controlling for known demographic and socioeconomic characteristics.
3 Developing measures of social sustainability and urban form at linked neighbourhood
level
Our empirical investigation of social sustainability is focused on fifteen case-study
areas within five medium-sized British cities. These areas (typically containing some
2000 households) were chosen to reflect a diversity of typical British urban forms, with
varying ages, types, and tenures of housing and sociodemographic profiles. In each city
one area was chosen close to the city centre, one towards the suburban periphery of the
main built-up area, and one in between. In each of these locational bands, tenure and
housing type varied considerably.
3.1 Household survey
In our view the principal source of evidence concerning the social acceptability of
different urban forms should be people themselves, particularly those living in the areas
in question. With this in mind, we designed a household-survey questionnaire to collect
data on social issues and other sustainability themes (for example, travel behaviour).
After consideration of different options (postal, drop-and-collect, telephone) we opted
for a self-completion postal method (with two reminders) and managed to achieve a
respectable 37% response rate (4381 responses). The survey responses reported below
have been reweighted using the 2001 Census to reflect the underlying demographic
profile of the population of each area, so countering possible differences in response
rates between demographic groups.
In designing the questionnaire we looked at a number of national and local surveys
covering similar topic areas and utilised or adapted some of the questions they con-
tained. This not only takes account of whether questions have been shown to work, but
also enables the wider benchmarking of our results. In general, we do not rely on a
single question to provide evidence about a given aspect of social sustainability but,
rather, draw on responses to a cluster of questions. In grouping responses together for
the composite social outcome measures we consider both their logical/linguistic inter-
pretation (`face validity') (Bryman, 2004) and patterns of correlations between the
responses across our sample.
3.2 Composite social outcome measures
We use the survey to generate ten composite measures which capture the different
aspects of social sustainability defined in section 2 above. Seven of the measures relate
to the sustainability of community, while three relate to access to and use of local
services.
Social sustainability and urban form: evidence from five British cities 2129
These composite outcome measures are expressed in an index form which is easy
to understand. For example, the `social interaction' measure is based on responses to
thirteen questions including whether respondents have friends in the neighbourhood,
know neighbours by name, look out for each other, chat, borrow things, and so forth
(1)
.
For each question, negative responses reduce the score from a neutral value, whereas
positive responses increase it (with neutral responses being possible in all cases).
Neutral responses across all the thirteen questions would result in a score of 100 on
the interaction index; all positive responses would score 200; and if all responses were
negative, the score would be zero. The resulting index scores are numbers in the range
0 ^ 200, but typically around or just above 100, for individuals. These are effectively
continuous variables and we can therefore compare mean values and variations
between and within groups, areas, or area types.
3.3 Measuring urban form
To establish relationships between these outcomes and urban form, we link the location
of sample households (addresses) to information about those locations, in particular
information about the small areas within which people live. It should be noted at this
point that the emphasis here is on the urban form of the residential neighbourhood
rather than, for example, the places where people work, shop, or spend their free time.
We have a choice about the spatial level at which urban form characteristics might
be attributed to individuals. This could be at any of the following levels: address of
individual house/property; street block; subarea; case-study neighbourhood; and city,
or even city ^ region.
For the analyses reported in this paper we rely mainly on linkage at the level of the
`subarea'. These subareas were defined using maps and local knowledge to identify natural
subdivisions via major boundary features. Each subarea is relatively homogenous in terms
of urban form, and there are six or seven in each case-study neighbourhood. This gives
a total of ninety-eight subareas, most of which have a population of around 400^ 500
households (a small number of subareas are mainly nonresidential and contain very
few households). Census and other neighbourhood data were apportioned to the
subareas from source units, census output areas, or other geographies. In addition to
a wide range of sociodemographic data, the census yields certain measures of urban
form (gross density, dwelling type, storey height).
The other key source of urban form measures is Ordnance Survey Mastermap,
which may be analysed using GIS to determine the proportion of land area attributable
to different elements (for example, residential buildings, gardens, greenspace, roads)
and associated ratios (eg percentage of greenspace, net dwelling density, average garden
size). Other urban form and quality measures derived from a site-inspection survey
were also incorporated (for example, building height, rundown areas, mixed use), along
with simple measures of access distance to city centres. A group of more sophisticated
network connectivity measures were derived from the Mastermap layers for roads and
pathways, utilising the technique of multiple centrality analysis (MCA), a development
of the space syntax concept (Cardillo et al, 2005)
(2)
. Other socioeconomic data
attached includes components from the indices of multiple deprivation (IMD), partic-
ularly overall deprivation, low income, and geographical barriers to accessing selected
facilities (effectively a proxy for more rural locations).
(1)
FulldetailsofthethesecompositesareprovidedintableA1onthejournalwebsiteathttp://
dx.doi.org/10.1068/a4184
(2)
ThesubsetofMCAmeasuresusedinthisanalysisispresentedintableA2onthejournal
websiteathttp://dx.doi.org/10.1068/a4184
2130 G Bramley, N Dempsey, S Power, C Brown, D Watkins
In addition to the array of neighbourhood (mainly subarea) urban form attributes,
we also include in our analysis certain indicators at the individual household/address
level, particularly housing type, floor (storey) height, and access to a garden.
A key issue in interpreting evidence on the association between social outcomes
and urban form is that of untangling `real' and potentially `causal' associations from
what may be apparent, fortuitous, or ambiguous relationships, given the complexities
of urban life and the different kinds of relationships which may be at work. Simply
showing that, in a two-way table or correlation, there is an apparent (negative)
relationship between, say, density and neighbourhood satisfaction, does not establish
that there really is such a relationship, let alone that it is causal. It is essential to take
account of (control for) the influence of other relevant variables, that is, other physical,
social, economic, or demographic factors which may also influence neighbourhood
satisfaction. For example, older people may tend to answer satisfaction surveys in
more positive ways than do younger people, and at the same time older people may
be more likely to live in low-density suburbs. To take another example, neighbourhoods
with concentrations of poor people tend to exhibit a greater incidence of certain social
problems which affect neighbourhood satisfaction; such neighbourhoods are also often
higher density areas.
3.4 Modelling outcomes
Although we use simple descriptive tables to show the patterns between different
forms, our main analysis and conclusions rely more on statistical modelling. Multiple
regression analysis provides the standard tool for establishing the direction and
strength of relationships while simultaneously taking account of the relationships
with other variables. Initial exploratory analyses were undertaken using ordinary least
squares (OLS) regression in the SPSS package to arrive at a provisional model for each
outcome. For the final models, however, we took account of the multilevel structure of
the data, recognizing that, while some variables were at the individual level, others
were at the higher subarea level, with the whole dataset effectively clustered by subarea
(Snijders and Bosker, 1999). The models were reestimated in more parsimonious forms
using a generalised linear model with random effects, implemented using the Stata
package (xtreg command). This approach has three main advantages: (a) robust stan-
dard errors and significance levels are generated for the explanatory variables, allowing
for the lesser degrees of freedom at subarea level, leading to more reliable inferences
about individual relationships; (b) as a byproduct, a number of variables of marginal
(nonrobust) significance could be dropped, yielding simpler, more parsimonious mod-
els; (c) the intercept (constant) in the model is allowed to vary between subareas,
reflecting random variations in outcome (not fully captured by explanatory variables)
at this level. It should also be added that using multilevel data helps to overcome the
problem of potentially misleading ecological correlations in traditional cross-sectional
regressions using grouped area-level data.
Although this approach goes beyond ordinary linear regression, it still represents a
relatively conservative modelling strategy. We do not systematically explore nonlinear
and interaction terms between variables, nor the possibility of `random slopes' (that is,
different sensitivities to particular variables in different subareas), although one or two
individual nonlinear versions of variables are tested (density, and distance from the
central business district. Such exploration could feature in future research. However,
it is important to note that, because of the range of different urban form indicators
used in the models, the overall `urban form effect' generally turns out to be nonlinear
when plotted against particular measures like density, as shown below.
Social sustainability and urban form: evidence from five British cities 2131
4 Empirical results
4.1 Descriptive profiles
Table 1 presents data on the urban form and access characteristics of the sample
neighbourhoods, breaking these down by location (inner, middle, and outer) and
gross residential density bands. It should be emphasised here that density measures
in the first two columns are the sample-weighted means of subarea-level figures
(the average of the localised densities at which people actually live), which are
systematically higher than conventional averages calculated over larger areas
(3)
.
The third column shows the simple gross density calculated at the larger case-study
neighbourhood level; these figures are typically much lower (half or less) than the
sample-weighted (or population-weighted) figures.
There is a strong gradient in terms of net dwelling density from inner (265) to outer
(96) locations but, rather surprisingly, little gradient in gross density (from 63 to 50).
Tabl e 1 . Urban form measures by location and density.
Net Gross Gross case- Distance Detached Terraced
dwelling dwelling study area from CBD dwelling dwelling
density density density (km) (%) (%)
nddenssa
a
gddenssa
a
gddenscs dist150k DwelDetach DwelTerr
Location
inner 264.8 63.0 29.1 1.0 1.2 12.5
middle 106.0 50.8 27.6 2.5 6.6 36.0
outer 96.2 49.6 16.4 6.7 20.9 17.1
total 158.1 54.7 24.4 3.4 9.4 21.6
Gross density (dwellings per hectare)
<20 93.3 14.3 20.2 4.0 19.3 11.2
20 ± 40 122.8 27.4 18.2 4.2 14.2 11.6
40 ± 70 150.7 52.6 25.8 2.5 4.1 26.7
>70 255.4 118.1 33.0 3.0 3.6 34.2
total 158.1 54.7 24.4 3.4 9.4 21.6
Flats Buildings Average Gardens Green Nonresidential
(%) >4 storeys garden (% area) space (%) and mixed
(%) size (ha) use (%)
DwelFlat pov4stor avgdnsiz presgdn pgreen pnrmix
Location
inner 81.1 61.6 0.003 10.5 26.6 16.2
middle 42.0 20.3 0.013 34.6 15.3 4.6
outer 19.4 2.1 0.024 45.6 18.2 1.6
total 48.2 28.7 0.013 29.8 20.2 7.7
Gross density (dwellings per hectare)
<20 48.5 29.9 0.024 28.8 25.7 8.6
20 ± 40 47.2 27.3 0.016 31.0 23.3 10.4
40 ± 70 50.1 30.5 0.009 28.2 16.8 7.6
>70 46.8 27.3 0.008 31.3 17.2 3.9
total 48.2 28.7 0.013 29.8 20.2 7.7
a
Gross density (gddenssa) is the sample-weighted mean of subarea-level gross dwelling
densities; case-study area density (gddenscs) is the simple ratio of dwellings to the total area of
whole case-study areas.
(3)
Our net density figures are also higher than some others because the denominator is solely
residential building footprint and garden plot areas, without any allowance made for streets or
pavements, as this can be more readily calculated from Mastermap.
2132 G Bramley, N Dempsey, S Power, C Brown, D Watkins
Gross and net densities do not necessarily move closely in step, because of the varying
presence of nonresidential land uses and undeveloped land. Our data indicate that,
in these cities, quite a lot of people are living at quite high net densities, even though
they may be in `outer' (suburban) locations, and even though the apparent gross density
is quite low, particularly when measured for larger zones (such as our case-study
neighbourhoods).
These are relatively compact cities and the average distance to the city centre is still
less than 7 km even in the `outer' areas. Nevertheless, there is a marked transition in
house-type mix from 81% flats and only 1% detached in the inner areas to 19% flats
and 21% detached in the outer areas. Terraced houses are most prevalent in middle-
ring areas and in areas with relatively high gross densities. More than three fifths of
buildings in our inner areas are four or more storeys in height, a figure which falls to
only 2% in the outer areas. Unsurprisingly, average garden size rises with distance from
the centre and as density falls: gardens represent only 10% of the land area in inner
areas but 47% of land in the outer areas. Greenspace, however, seems to have a slightly
higher share of land in the inner areas than in the outer ones, although the absolute
amount is smaller. Nonresidential and mixed uses are mainly found in inner and
middle areas.
Tabl e 2 . Social sustainability outcomes by location and density.
Pride, Social Safety Environmental Satisfaction
attached interaction quality with home
nhpride nhinter nhsafe nhenvir homesat
Location
inner 92.5 97.3 123.4 100.4 117.3
middle 115.7 122.5 133.0 120.2 130.9
outer 118.4 129.1 126.9 129.8 139.7
total 108.5 115.9 127.6 116.4 129.0
Gross density (dwellings per hectare)
<20 116.3 118.4 134.4 120.6 139.0
20 ± 40 112.7 116.4 131.4 121.1 131.3
40 ± 70 109.0 117.3 127.2 113.9 125.7
>70 97.3 111.7 119.1 111.2 123.6
total 108.5 115.9 127.6 116.4 129.0
Stable Participation Use of Frequency of Frequency of
versus in groups neighbourhood use of utility use of leisure
mobile facilities/services services services
stable grppart usenfac utilfreq leisfreq
Location
inner 109.7 94.2 108.5 236.2 103.5
middle 127.6 101.5 103.0 223.3 81.1
outer 147.1 104.6 88.0 188.1 51.2
total 127.7 100.0 100.0 216.3 79.1
Gross density (dwellings per hectare)
<20 132.4 105.4 102.4 207.4 81.7
20 ± 40 132.1 106.0 98.2 211.0 82.9
40 ± 70 121.0 97.5 105.7 236.7 85.8
>70 128.1 92.5 92.7 202.1 63.5
total 127.7 100.0 99.9 216.2 79.0
Social sustainability and urban form: evidence from five British cities 2133
Table 2 shows average scores for the ten composite social sustainability performance
measures for the same breakdown of sample areas in terms of location and gross
density. For all indicators higher scores indicate a more favourable outcome. These
descriptive profiles, of course, simply show the raw bivariate relationship with these
urban form factors, without any attempt to control for other influences.
For three of the outcomes there is a common pattern of ranking, broadly in line
with expectations, whereby outer areas score better than inner areas and less dense
areas score better than denser areas
ö
this applies to pride/attachment, social inter-
action, and satisfaction with the home. The differences generally seem greater between
the locational bands than between the gross density bands
For four of the outcomes there is a modified or attenuated version of this pattern.
In relation to safety, while inner areas are worst, middle rather than outer areas are
best (within our sample of cases).
(4)
For environmental quality and also for participa-
tion in groups, medium density [20^ 40 dwellings per hectare (DPH) gross] is slightly
better than the lowest density bands. For stability versus mobility, both low-density and
medium-density areas score similarly and better than average, while medium ^ high-
density areas (40^ 70 DPH) scores lowest. These patterns suggest possible non-linear
relationships with density.
For the remaining three outcomes, those relating to usage of local services, the
pattern is different; broadly as expected although with some deviation. For each of
these measures, inner areas score better than middle areas which score better than
outer areas, as expected on grounds of accessibility and availability of services.
Frequency of use of leisure services shows the greatest difference between inner and
outer areas of all of the indicators. This perhaps confirms the significance of leisure,
entertainment, and cultural amenities to an `urban way of life'. The relationship with
density generally shows a rise with density up to the medium-high band (40 ^ 70 DPH),
but then a marked falling off in the highest band.
4.2 Modelling results overall
The full regression model details for the ten social outcomes are not shown in the body
of this paper to save space but are provided on the journal website at http://dx.doi.org/
10.1068/a4184 (table A5). Altogether, fifty-four variables are included as significant for
at least one outcome, with about half of these being individual household-level vari-
ables and half being subarea attributes. Typically, individual outcome models have
fifteen to twenty significant variables, broken down into the following four categories:
D
ö
demographic; S
ö
socioeconomic; A
ö
access/location; U
ö
urban form.
(5)
The overall model performance (R
2
statistics) can be partitioned into the propor-
tions of variance explained `between'groups (that is, subareas) and `within'groups (that
is, between individuals within subareas) (Stata, 2005, pages 282 ^ 308). The overall
proportion of variance explained (between individual observations) is typically around
0.2 to 0.3, not an uncommon figure in microanalyses of social phenomona. However,
it is noteworthy that for most of the outcomes the level of explanation between the
(92 ^ 94) subareas is much higher, at around 0.60 ^ 0.85. So, compared with with a
simpler, more traditional cross-sectional regression of area-level data, the fit of the
models is quite high. In other words, we are explaining most of the variation between
different neighbourhoods and, since urban form is essentially an area phenomenon,
this is very encouraging.
(4)
There are significant peripheral council housing estates in two of our outer areas, which may
affect this finding.
(5)
Size of dwelling attributes (for example, rooms), are treated as a measure of material living
standard, associated with income/wealth, and hence placed in the socioeconomic category rather
than the urban form category.
2134 G Bramley, N Dempsey, S Power, C Brown, D Watkins
4.3 Specific relationships
In commenting on each outcome in turn, we focus primarily on access and urban
form effects, but also highlight noteworthy effects among the demographic and
socioeconomic control variables.
`Neighbourhood pride and attachment' is clearly related to having a (larger) garden,
to lower gross density, and to being in a more rural location. There is a marginal
negative effect from the MCA `betweenness' score as well. These results are in line
with the predominant picture in the literature, and apply after controlling for a range
of demographic effects (particularly age) and socioeconomic effects (tenure, working,
poverty, and smaller dwellings
ö
the last two being negative).
`Social interaction' (including `friendliness') is similar in some ways to `neighbour-
hood pride and attachment', but different in others. Families and longer residence are
associated with more social interaction, with younger and single-person households
being less engaged. There is still clearly a relationship with having a garden and, in
addition, living above the ground floor shows up as negative. There is no apparent
relationship with either of the density measures. However, for this outcome, living
further out from the city centre (log distance) is negative and the MCA accessibility
measure is positive. Two further noteworthy associations are that terraced housing
areas score positively and that there is also a positive association with neighbourhood
greenspace. We believe these findings are consistent with some of the literature, and
with common sense. Gardens are conducive to interaction, both as a locus for social
activity but also as a relatively safe, `semiprivate' outdoor space which provides oppor-
tunities to interact with neighbours and passers-by. High overall densities have the
problems of anonymity and psychic overstimulation, and higher storey flats provide
fewer opportunities for comfortable meeting on the street or near the entrance to the
home. Terraced housing is the best for this, because the houses are close enough for
people to bump into each other or to talk across the wall or fence; in low density
detached suburbia, people are more likely to go from their house door to their car and
then drive away. In addition, in areas of higher general accessibility it is easier for
people to meet up, and more interaction can be created indirectly through use of local
services or public transport.
The `safety' outcome has a negative association with gross density and with one
MCA accessibility measure, and a positive association with the proportion of residen-
tial garden area. These associations are in line with what we would expect from the
literature, suggesting that `suburbs' are safer (or are perceived to be safer). However,
safety actually falls with distance from the central business district (CBD) in this
sample, when other effects are controlled for; there is also a negative association with
the MCA variable msteff, which indicates that areas which have a more elemental
network (with less redundancy, that is, cul de sacs versus grids) are, or feel, less safe.
Safety is generally greater for males, younger people, private renters, the better off, and
people living in less poor areas.
`Neighbourhood environmental quality' has a negative association with net resi-
dential density, terraced housing, higher buildings, and being on through routes, while
being positively associated with more rural locations. Social renting, poverty, and lower
income are clearly associated with worse environments, whereas the relationships with
age and length of residence are more complex. These findings are consistent with some
previous literature, but we would suggest that experience of environmental quality is
not just a function of physical design features but also of the care and maintenance
services in place. `Poor' and `inner' areas may need more of these services to bring
standards up to a decent level (Fisher and Bramley, 2006; Hastings et al, 2005).
Social sustainability and urban form: evidence from five British cities 2135
`Satisfaction with the home' shows strong relationships with (mainly) individual-level
housing type and associated urban form attributes, as expected. Satisfaction is clearly
associated with having a garden, with bungalows and detached houses, and with being
in areas with larger gardens, while it is negatively associated with terraced houses.
Satisfaction is clearly greater for households with more rooms, and less for social
renters (who tend to have less choice). Older people are more satisfied; younger, larger,
lone-parent, and nonwhite households are less satisfied.
`Residential stability' is seen in some literature as a key to neighbourhood social
capital, but it is clearly strongly related to lifestage demography (age, family status, and
family size) and tenure (private renting). Lack of a garden is again negative for this
outcome. Net residential density has a negative effect on stability, but this is nonlinear
(the quadratic term means that this negative effect declines to a minimum around
250 DPH). Nonresidential and mixed uses are also negative for stability; this may reflect
the environmental disturbance effect of these other nearby uses, but it may also partly
reflect any lifestage selection effects not fully controlled for by the demographic and
socioeconomic variables.
`Participation in groups' is also seen in some literature as a key proxy for social capital.
However, in this study we have found that it is the outcome for which we have the poorest
predictive power in the models
ö
even at the area level (the `between' subareas R
2
is only
0.234). This may be partly because we chose to use a combination of participation in
activities at neighbourhood and city levels in this index. The only two urban form or
access-related variables included in this model are `not having a garden' (negative) and
`nonresidential and mixed uses' (positive). This last relationship is consistent with `new
urbanist' views that mixed use promotes community. The former may be an indirect
selection effect, proxying people without a longer term commitment to place. Families,
larger households, non-elderly households and nonwhite households participate more
in collective activities; private renters and higher income households participate less.
The associations with subarea socioeconomic status are weak.
The overall index for `use of neighbourhood facilities' shows relationships with a
number of access and urban form variables, and the direction of these relationships is,
as expected, rather different from that of many of those outlined above. Use of local
services is less at greater distances from the city centre and more where the MCA
accessibility indicator is greater. There is a negative relationship with the proportion of
residential gardens and greenspace, with residential gardens in particular a general
`suburban' proxy. These relationships are as expected from the literature, but the
negative relationship with net residential density is not (although, as shown in table 2,
the relationship with density is a bit ambiguous). Where respondents find it less easy
to access public transport on foot and where services are less frequent, there is a
significant negative association with use of neighbourhood facilities. In other words,
better public transport goes with more use of local services, and vice versa. Older
and family/larger households are more likely to use neighbourhood services, whereas
younger and single-person households are less likely use them. Renters are more likely,
and working households less likely to use such services. There are negative associations
both with area deprivation and with `rundown' areas, suggesting that for these places
there may be safety and quality issues which deter use of services.
The last two outcome measures disaggregate services into two types: everyday
`utility' services and more discretionary `leisure'and cultural services. They also measure
usage by annual frequency (estimated from banded data). There are some detailed
differences when these variables are used, but broad consistency.
2136 G Bramley, N Dempsey, S Power, C Brown, D Watkins
4.4 Urban form, access, and other effects
In order to get a fair picture of the effects of urban form on social outcomes it is
necessary to calculate the effects of all of the relevant variables included in each model.
We have therefore constructed `component' measures which are the product of the
value of each included variable within a category (for example, urban form) and its
estimated coefficient from the model, with a constant term to ensure that each com-
ponent is centred on the same overall mean score. Components were calculated in this
way for each outcome by each of four categories: urban form, location/access, demog-
raphy; and socioeconomics. If we take a particular measure of urban form (for example,
net density) as a representative measure, we can then plot these components against this
to show the general pattern of urban form and other effects, which go to make up the
overall outcome pattern. Figure 1 illustrates this for three of our outcome measures:
neighbourhood pride and attachment, social interaction, and use of neighbourhood
services. As anticipated at the end of section 3, the urban form component as plotted in
Use of facilities score Neighbourhood-interaction score Pride and attachment score
160
140
120
100
80
60
140
130
120
110
100
90
80
120
110
100
90
80
Overall
Urban form
Location/access
Demographic
Socioeconomic
(a)
(b)
(c)
0 100 200 300 400
Net dwelling density (dwellings per hectare)
Figure 1. (a) Pride and attachment by density, (b) social interaction by density, (c) use of neighbour-
hood facilities by density.
Social sustainability and urban form: evidence from five British cities 2137
this way does not in general have a linear relationship with density
ö
despite the fact that
the specific relationships underlying this component are generally linear.
Figure 1(a) shows how the overall index for neighbourhood pride and attachment
drops as density rises, from nearly 150 in the lowest density band to just under 90 at a
net density level of 150 DPH (net), with a more or less constant score at higher levels.
The figure also shows how urban form and other components contribute to this out-
come. Urban form also has a negative but diminishing (that is, nonlinear) effect on
pride and attachment, but does not account for all of the overall decline. Demographic
factors also play a part, along with a slight effect from location/access and a very sharp
negative effect from socioeconomic factors in the 100^ 150 DPH range. This figure
illustrates a common pattern across quite a number of the outcomes, whereby appar-
ently quite strong negative associations between density and social outcomes are
actually somewhat attenuated when allowance is made for other factors.
Figure 1(b) presents a similar picture for social interaction. Here, although there is
a similar overall relationship with density (negative, diminishing), the role of the differ-
ent categories of variable is quite different. Social interaction increases up to a density
level of 100 DPH (net), before dropping off somewhat at higher levels. The effect of
location/access is similar, raising interaction up to a density level of 150 DPH, then
falling slightly. However, these effects are offset by quite negative effects in the range up
to 100 ^ 150 DPH both from demographic and from socioeconomic factors. This figure
helps to underline the messages from the specific variable effects, that a number of
features associated with medium density are most productive of social interaction.
Figure 1(c) presents a similar analysis for use of neighbourhood facilities, where the
pattern is, as expected, rather different. Here, the overall score rises strongly to a peak
around 150 DPH, then falls off before rising somewhat in the very highest density
band. The relationship with density up to the peak and beyond is strongly driven
both by urban form and by location/access effects. These are not offset by moderate
countervailing demographic and socioeconomic effects.
Reviewing graphs of this kind across all the outcomes suggests that our ten mea-
sures of social sustainability fall into three main groups in terms of their performance
across the urban form spectrum. These three groups favour lower density (pride/
attachment, safety, environment, home satisfaction, stability), medium density (interaction
and participation), and higher density (access to and use of local services), respectively.
5 Concluding discussion
In this paper we consider the relationship between urban form and aspects of the
social sustainability of communities, in the context of existing medium-sized British
cities. The measurement of density itself is not simple, and our findings suggest the
need to take account of housing-type mix, the role of gardens and greenspace, land-use
mix, and network-connectivity characteristics.
There are, nevertheless, certain typical patterns in the relationship of social out-
comes with density and location. Broadly, social outcomes relating to attachment,
satisfaction, safety, and environment are more positive at lower densities and in less
central locations, while outcomes relating to use of local services display the opposite
tendency. Even these simple bivariate patterns suggest some nonlinearities, and also
that some outcomes are more strongly related to urban form than others.
Multilevel regression models linking these outcomes to a wide range of indi-
vidual and small area attributes typically explain 25% of total individual variance,
but in most cases 60%^ 85% of between-subarea variance. The regression models
serve to control for the potentially confounding effects of demography and socio-
economic status, and it turns out that allowing for these controls significantly
2138 G Bramley, N Dempsey, S Power, C Brown, D Watkins
modifies relationships between many of the social outcomes and urban form and/or
location/access.
Firstly, outcomes relating to neighbourhood pride and attachment, stability, safety,
environmental quality, and home satisfaction all display a negative, nonlinear relationship
with density. Secondly, outcomes relating to social interaction and group participation tend
to improve as density rises up to a medium level, and then fall off at higher levels. Thirdly,
outcomes relating to the use of local services are broadly positively related to density.
This third group represents the `equity' aspect of social sustainability, whereas the
previous two groups represent the `community' aspect. However, greater use of local
services also contributes to greater sustainability in terms of travel and transport,
which itself displays a similar pattern over density.
The outcomes with the strongest negative relationship with density are home
satisfaction and safety, both important factors influencing residential location choices,
suggesting that housing-market choices are still likely to favour lower density options
where available. The outcome which is least related to urban form is participation
in collective community activities, frequently used as a marker of `social capital'.
Some of the specific variable effects are particularly interesting in their own right:
for example, the positive role of gardens, not just in relation to environmental benefits
but also in terms of outcomes such as social interaction.
The findings suggest that there is a kind of `density divide' at around 100 ^ 140
DPM (net). The main relationships identified apply up to this level; beyond this density,
relationships are either different or not apparent at all. We would be reluctant to draw
any strong conclusions about the merits or otherwise of `superdensity' from this
evidence.
(6)
The numbers of observations in these higher reaches of the density spec-
trum are rather sparse, and the building forms involved may not be representative of
contemporary `super density' design proposals.
It is clear that, when other dimensions of sustainability are brought into account,
the patterns of outcomes on these may be different from those we have found for the
social dimensions. Policy assessment is likely to involve trade-offs. The obvious case
will be transport and travel, and indeed we can demonstrate from this dataset that
sustainable travel is likely to favour higher densities and compact forms (see also
Ferguson and Woods, 2009). This clearly conflicts with, or must be traded off against,
the `community'/'attachment' aspects of social sustainability, but is actually rather
similar to the `equity' aspect of social sustainability, at least insofar as this is captured
by measures of use of local services.
The most important policy implication of these findings is that care is needed in
planning new urban developments or redevelopments if these are to be socially accept-
able and successful communities. An exclusive emphasis on high density, particularly
if this takes the form of apartment accommodation with little provision of gardens,
is unlikely to produce happy, well-functioning communities. Compromises between the
arguments (particularly from the sustainable transport perspective) for high density
and the social and quality of life considerations will be needed. The socioeconomic and
demographic mix of communities is also shown in this study to be very significant
for social outcomes, implying that policy should also promote mix and balance in this
respect.
In planning for future urban developments, or the reshaping of existing cities, it is
also clear that `one size does not fit all' and that there will be a need to provide a range
of environments for different groups in different locations. There is a need for `whole
system' simulations, allowing for the different locational and lifestyle choices which
(6)
See Design for London (2007) for a recent perspective on superdensity.
Social sustainability and urban form: evidence from five British cities 2139
different groups will make and the likely outcomes which result. The data generated
in this study can contribute to the prediction of what those outcomes would be likely
to be. However, it would be valuable to extend the database to take account of
behaviour and outcomes in a broader range of urban form and settlement contexts.
Such work should clearly also consider social and demographic mix of population,
alongside urban form, when informing the planning of new communities.
Acknowledgements. The research was carried out as part of `City-Form: the Sustainable Urban
Form Consortium' (http://www.city-form.org) funded by the Engineering and Physical Sciences
Research Council (Grant GR/S20529/01) under its Sustainable Urban Environment programme.
References
ACPO, undated, ``Secured by design: the official police flagship for security and designing out
crime'', Association of Chief Police Officers Crime Prevention Initiatives Ltd, 1st Floor,
10 Victoria Street, London SW1H 0NN
Alexander C, Ishikawa S, Silverstein M, Jacobson M, Fiksdahl-King I, Angel S, 1977 A Pattern
Language: Towns, Buildings, Construction (Oxford University Press, Oxford)
Arnott R, 1998, ```Economic theory and the spatial mismatch hypothesis'' Urban Studies 35
1171 ^ 1185
Audirac I, Zifou, M, 1989,``Urban development issues: what is controversial in urban sprawl?
An annotated bibliography of often overlooked sources'', Bureau of Economic and Business
Research, College of Business Administration, University of Florida
Bailey N, Livingstone M, 2007 Population Turnover and Area Deprivation research report, Joseph
Rowntree Foundation,York
Barton H (Ed.), 2000 Su stainable Communities: The Potential for Eco-neighbourhood s (Earthscan,
London)
Bramley G, Morgan J, 2003, ``Building competitiveness and cohesion: the role of new house
building in central Scotland's cities'' Housing Studies 18 447 ^ 471
Bramley, G, Power S, 2009, ``Urban form and social sustainability: the role of density and housing
type'' Environment and Planning B: Planning and Design 36 30 ^ 48
Bramley G, Brown C, Power S, Dempsey N, 2006, ``What is sustainability and how do existing
urban forms perform in nurturing it?'', paper presented at the Planning Research Conference,
University College London; copy available from the authors
Breheny M, 1992a Sustainable Development and Urban Form (Pion, London)
Breheny M, 1992b, ``The contradictions of the compact city: a review'', in Sustainable Development
and Urban Form (Pion, London) pp 138 ^159
Bridge G, 2002,``The neighbourhood and social networks'', CNR, Paper 4, April,
http://www.neighbourhoodcentre.org.uk
Bryman A, 2004, Social Research Methods (Oxford University Press, Oxford)
Bunker, R, 1985, ``Urban consolidation and Australian cities''Built Environment 11 83 ^ 96
Burton E, 2000a, ``The compact city: just or just compact? A preliminary analysis'' Urban Studies
37 1969 ^ 2001
Burton E, 2000b,``The potential of the compact city for promoting social equity'', in Achieving
Sustainable Urban Form Eds K Williams, E Burton, M Jenks (E&FN Spon, London)
pp 19 ^ 29
Calthorpe P, 1993 The Next American Metropolis: Ecology, Community and the American Dream
(Princeton Architectural Press, New York)
Cardillo A, Scellato S, Latora V, Porta S, 2005 Structural Properties of Planar Graphs of Urban
Street Patterns University of Catania, Sicily
CEC, 1990 Green Paper on the Urban Environment (Commission of the European Communities,
Brussels)
CNU, 2004 Charter of New Urbanism Congress of the New Urbanism, Chicago, IL,
http://www.cnu.org/about/index.cfm
Collie M, 1990,``The case for urban consolidation'' Australian Planner 28 26 ^ 33
Dempsey N, Bramley G, Power S, Brown C, 2009,``The social dimension of sustainable development:
defining urban social sustainability'' Sustainable Development 17 forthcoming
Design for London, 2007 Superdensity: Recommendations for Living at Superdensity Design for
Homes, 26 Store Street, London WC1E 7BT, http://www.designforhomes.org
DETR, 1999a Towards an Urban Renaissance
ö
Final Report of the Urban Task Force Department
for Environment, Transport and the Region (E&FN Spon, London)
2140 G Bramley, N Dempsey, S Power, C Brown, D Watkins
DETR, 1999b A Better Quality of Life Department for Environment,Transport and the Regions,
London
Diamond H L, Noonan P F, 1996 Land Use in America (Island Press,Washington, DC)
Duany A, Plater-Zyberk E, 2001 The Rise of Sprawl and the Decline of the American Dream
(North Point Press, New York)
Ewing R,1997,``Is Los Angeles style sprawl desirable?''Journ al of the Ame rican Planning Association
63 107 ^ 126
Ferguson N,Woods L, 2009,``Urban form and travel behaviour'', in Dimensions of the Sustainable
City Eds M Jenks, C Jones (Springer, New York) forthcoming
Fisher T, Bramley G, 2006,```Social exclusion and local services'', in Poverty and Social Exclusion
in Britain Eds C Pantazis, D Gordon, R Levitas (Policy Press, Bristol) pp ...
Forrest R, Kearns A, 1999 Joined-up Places? Social Cohesion and Neighbourhood Regeneration
York Publishing Services for the Joseph Rowntree Foundation,York
Forrest R, Kearns A, 2001, ``Social cohesion, social capital and the neighbourhood''Urban Studies
38 2125 ^ 2143
Freeman L, 2001, ``The effects of sprawl on neighbourhood social ties'' Journal of the American
Planning Association 67 69 ^ 77
Glynn T,1981,``Psychological sense of c ommunity: measurement and application'' Human Relations
34 789^818
Gordon P, Richardson H, 1997,``Are compact cities a desirable planning goal?''Journal of the
American Planning Association 63 95 ^ 106
Hastings A, Flint J, McKenzie C, Mills C, 2005 Cleaning Up Neighbourhoods: Environmental
Problems and Service Provision in Deprived Areas (Policy Press, Bristol)
Haughton G, Hunter C, 1994 Sustainable Cities (Jessica Kingsley, London)
HM Government, 2005 Securing the Future Cm 6467 (The Stationery Office, London)
Houston D, 2005, ``Employability, skills mismatch and spatial mismatch in metropolitan labour
markets'' Urban Studies 42 221 ^ 243
Ihlandfeldt, K, Sjoquist D, 1998,``The spatial mismatch hypothesis: a review of recent studies
and their implications for welfare reform'' Housing Policy Debate 9849 ^ 892
Jacobs J, 1961 The Death and Life of Great American Cities (Penguin Books, Harmondsworth,
Middx)
Jenks M, Burton E,Williams K (Eds),1996 The Compact City: A Sustainable Urban Form?
(E&FN Spon, London)
Katz P, 1994 The New Urbanism: Toward an Architecture of Community (McGraw-Hill, New York)
Nasar J, Julian D, 1995, ``The psychological sense of community in the neighbourhood'' Journal
of the American Planning Association 61 178 ^ 184
Nash V, Christie I, 2003 Making Sense of Community Institute for Public Policy Research, London
Nelessen A C, 1994 Visions for a New American Dream: Process, Principles and an Ordinance to
Plan and Design Small Urban Communities (American Planning Association, Chicago, IL)
ODPM, 2003 Sustainable Com munities: Build ing for the Future Office of the Deputy Prime Minister
(The Stationery Office, London)
ODPM/Home Office, 2004 Safer Places: The Planning System and Crime Prevention (Home Office,
London)
Pantazis C, Gordon D, Levitas R, 2006 Poverty and Social Exclusion in Britain (Policy Press,
Bristol)
Pierson J, 2002 Tackling Social Exclusion (Routledge, London)
Putnam R D,1993 Making Democracy Work: Civic Traditions in Modern Italy (Princeton University
Press, Princeton, NJ)
Shaftoe, H, 2000,``Community safety and actual neighbourhoods'', in Sustainable Communities:
The Potential for Eco-neighbourhoods Ed. H Barton (Earthscan, London) pp 230 ^ 2 45
Shore W B, 1995,`` `Recentralization': the single answer to more than a dozen United States problems
and a major answer to poverty'' Journal of the American Planning Association 61 496 ^ 503
Simmel G, 1995 [1903], ``The metropolis and mental life'', in Metropolis: Center and Symbol of
our Times Ed. P Kasnitz (Macmillan,Basingstoke, Hants) pp 30 ^ 45
Snijders T, Bosker R, 1999 Multilevel Analysis: An Introduction to Basic and Advanced Level
Multilevel Modelling (Sage, London)
Stata, 2005 Stata Release 9: Longitudinal and Panel Data (Stata Press, College Station, TX)
Talen E, 1999, ``Sense of community and neighbourhood form: an assessment of the social doctrine
of new urbanism'' Urban Studies 36 1361 ^ 1379
Social sustainability and urban form: evidence from five British cities 2141
TCRP, 1998, ``The costs of sprawl revisited'', Report 39, Transit Cooperative Research Program,
Transportation Research Board, Washington, DC
Williams K, 2000,``Does intensifying cities make them more sustainable'', in Achieving Sustainable
Urban Forms Eds K Williams, E Burton, M Jenks (E&FN Spon, London) pp 30 ^ 45
Wilson J, 1985 Thinking about Crime (Random House, NewYork)
Wirth L, 1938,``Urbanism as a way of life''American Journal of Sociology 44 1^24
Woolley H, 2002, ``Inclusive op en spac es'' Landscape Design number 310, pp 43 ^ 45
Worpole K, 2003, ``The social dynamic'', in Urban Villages and the Making of Communities
Ed. P Neal (Spon, London) pp 118 ^ 131
ß 2009 Pion Ltd and its Licensors
2142 G Bramley, N Dempsey, S Power, C Brown, D Watkins
Conditions of use. This article may be downloaded from the E
&
P website for personal research
by members of subscribing organisations. This PDF may not be placed on any website (or other
online distribution system) without permission of the publisher.
