The Density Debate in Urban Research: An ... - Wiley Online Library

bs_bs_banner

The Density Debate in Urban Research: An Alternative Approach to Representing Urban Structure and Form MICHAEL GROSVENOR* and PHILLIP O’NEILL Urban Research Centre, University of Western Sydney, Locked Bag 1797, Penrith, NSW 2751, Australia. *Corresponding author. Email: [email protected] Received 11 June 2014; Revised 11 August 2014; Accepted 27 August 2014

Abstract Global aspirations for urban sustainability coincide with debate over urban form. Much of the debate centres on the merit of increasing residential densities across metropolitan areas. We argue that this focus on density is problematic. Specifically, we question whether urban density is a sufficient proxy for representing urban structure and form across large metropolitan areas, as was promulgated famously by Newman and Kenworthy some time ago. Our concern is that density tells us little about a neighbourhood’s location, accessibility and design characters. Moreover, the focus on density may well be contributing to poorly located and designed development across metropolitan areas. Here we offer an alternative approach using a typology of urban structure and form to explore urban sustainability potential, specifically through three variables: transport mode choice, energy consumption, and water consumption. Our results are significant and suggest new approaches to urban form and structure in urban planning frameworks. KEY WORDS urban structure; urban form; density; sustainability

Introduction There is a widely held belief that urban design and planning are important influences in achieving sustainable outcomes. This belief has more or less held sway since the release of the United Nation’s seminal document, Our Common Future: The Brundtland Report, in 1987. The Brundtland Report introduced the concept of sustainable development, defined as ‘development that meets the needs of the present without compromising the ability of future generations to meet their own needs’ (Brundtland Commission, 1987, 41). The report provided encouragement to government agencies around the world, including Australia, to revitalise support for environmental planning objectives (Gleeson et al., 2004; 442

Anderson et al. 1996; Forster, 2006). Metropolitan strategies, in particular, became ‘pivotal instruments in the application of the governance hypothesis to the achievement of urban sustainability’ (Gleeson et al., 2004, 350). In this context, urban consolidation became the prime planning objective in most Australian states and territories (Gleeson et al., 2004, 352). The Brundtland Report was the impetus for urban architectural, planning, and design practices around the world to be ‘green and global’. A resurgence of interest in compact city theories and policies was one consequence (Burgess, 2000, 9–10), with supporters placing significant value on the environmental benefits that containment and consolidation might bring, particularly Geographical Research • November 2014 • 52(4):442–458 doi: 10.1111/1745-5871.12084

M. Grosvenor and P. ONeill: The Density Debate in Urban Research

through lower energy consumption, especially transport-related energy use (see Perkins et al., 2009; Rickwood, 2009; Ewing and Fong, 2010; Newton, 2011; Talen, 2011). Yet not all agreed. Peter Hall suggested, ‘That like most fixed ideas this one has a small element of truth and a much larger element of myth’ (Hall, 1997: 214); and others like Troy (2004), Gray et al. (2008), and Neuman (2005) concurred. For Neuman (2005, 23), the major concern was the way outcome was being extrapolated crudely from an observation of urban form: Form, as biologists and geologists understand it, is an outcome of evolution. Form is a snapshot of process. It is a fixed condition at any point in time. Form, in and of itself, is not measurable in terms of sustainability. Asking whether a compact city, or any other form of the city, is sustainable is like asking whether the body is sustainable. The proper question is not if the body is sustainable, but rather, does the being that inhabits the body live sustainably? So while resource use and environmental impacts of private household consumption are actually the key aspects of sustainable development (Holden and Norland, 2005, 2145) rather than density configurations per se, the links between household consumption behaviours and the built environment are very much underexplored (Gray et al., 2008, 17); or, perhaps worse still, portrayed crudely. Neuman (2005) argues that the methodological premium in urban form research accorded to a single operational measure, namely population density, when measuring urban form differentiation tends to reduce a complex entity (the city) to one criterion (density), thereby constraining research understandings and biasing policy action. Taking on Neuman’s challenge for more nuanced research, we attempt a substantially different approach to measuring associations between urban structure and form and sustainability outcomes. Our interest is in whether an urban structure and form typology approach can better predict the sustainability characteristics of Australian urban areas. The paper has three parts. First, we discuss the weaknesses of using density as a primary indicator of desirable urban form. We then present an overview of how an urban structure and form typology can be devised and used to test the strength of association with key urban sustainability data. We use rich data sets and statistical analysis to assess the effectiveness © 2014 Institute of Australian Geographers

443

of our typology of urban structure and form, before making some brief conclusions about policy and further research. The problem with density measures Density calculations are commonly used in urban research as a way of quantifying differences in urban form type, typically to do little more than represent an area as either high density or low density. Churchman (1999) provides a useful overview and assessment of common density measures, pointing to dwellings per hectare, population per hectare, and people plus jobs per hectare as the most popular. But it is Newman and Kenworthy who have been most influential in planning discourse as a result of their use of density calculations and associations to promote high density as the most sustainable of urban forms (see especially Newman and Kenworthy, 1989; 1999; 2000). In their work, Newman and Kenworthy calculate average population densities for most major cities in North America, Australia, Europe, and Asia, and correlate these with car use and energy consumption. They show that car use and, as a result, energy consumption increase exponentially once densities fall below around 30 persons per hectare. Newman and Kenworthy showed that Sydney’s car use, for example, was predictably higher in its outer urban areas as densities fell below the 30 dwellings per hectare benchmark. For Newman and Kenworthy, then, density ratios can readily account for high car use in Australian and North American cities, and high public transport use in Europe and Asia. Newman and Kenworthy’s research results have garnered much support (Rickwood, 2009), with Mees (2009) suggesting that their widely used graph showing the relationship between density and car use has significantly influenced metropolitan planning strategies in promoting urban consolidation policies in order to increase public transport use and reduce car use. Yet Mees sees Newman and Kenworthy’s research as flawed. Mees shows that the urban area boundaries used by Newman and Kenworthy to calculate urban density are inconsistently defined across the chosen cities – for example, some areas include surrounding suburban areas and some do not – and, after recalculating for every North American and Australian city considered by Newman and Kenworthy, finds that density does not in fact have the influence on transport patterns and behaviours that Newman and Kenworthy assert.

444 But even if calculative practices are of a high standard, should density calculations be used at all to represent a certain urban form ideal? Certainly, practitioners of planning and design in Australia currently use dwellings per hectare metrics to translate the perceived desirability for higher densities into specific built form outcomes (Griffiths, 2009). Sydney’s Metropolitan Strategy 2005, for example, defines high-density housing as more than 60 dwellings per hectare and generally five storeys or more in height (in other words, high-rise apartments). Medium density is defined as 25 to 60 dwellings per hectare and not usually more than three or four storeys in height (walk-up apartments, townhouses, and terrace houses). Low density is below 25 dwellings per hectare and would comprise predominantly detached dwellings (Griffiths, 2009, 4). The problem with such categorisations is that different built form types can generate a variety of different population density categorisations. Using NSW Urban Design Advisory Service density guidelines, Griffiths (2009, 3) shows that the same number of people (168) can be housed in medium-density (three storey) apartments (69 dwellings/hectare and 2.6 people/ hectare) as can be housed in high-density (four to nine storey) apartments (141 dwellings/hectare at 1.4 people/hectare). In other words, it is easy to show that the same population density can be calculated for very different dwelling and built form outcomes. Griffiths (2009) and Bamford (2009) both conclude that that the assumed correlation between higher population densities and higher residential building density is misleading. Neuman (2005, 21) is critical of those researchers using density calculations in urban form research. He states that as a representation of urban form, average density ignores variations in density within aggregated areas and fails to reflect variations in built form, transport service levels, and community linkages. Several other authors (McLoughlin, 1991,; Forster, 2006; Bamford, 2007, Griffiths, 2009; Mees, 2009) are similarly concerned at the negative impact on Australian urban planning policy of the focus on density. This is not to say that increased population density is incapable of delivering environmental, economic, and social benefits (Newman and Kenworthy, 1989; Bramley et al., 2009; Perkins et al., 2009; Jones et al., 2010; Florida, 2012); but, as Rickwood and Glazebrook show (2009: 184), there are likely to be other contributing factors. Despite such warnings, an overzealous application of density measures in urban

Geographical Research • November 2014 • 52(4):442–458

research persists, resulting in crude urban consolidation policy responses in many of Australia’s largest cities, especially Sydney, typically involving multi-level apartment construction and roll-out seen by a broad planning fraternity as an action towards the greater good (Randolph, 2004; 2006; Randolph et al., 2007). The question then arises as to why density calculations are used at all to represent urban form and structure differences. The answer is simple: in a statistical sense, density calculations make it easier to quantify the strength of a causal relationship by calculating the linear regression (r2) of a dependent variable (e.g. public transport patronage or energy consumed) against an independent variable representing urban form, such as density. Yet the robustness of such analysis is undermined if the independent variable does not validly represent urban structure and form difference. Not only can the same density correspond to different built or urban form outcomes (see Griffiths, 2009, above), but densities calculated over aggregated, often large, urban areas disguise variations in land use, physical design, social characteristics, and ecological conditions (Neuman, 2005, 21). Figure 1 illustrates the problem. Two districts, an ‘Inner Area’ and a ‘Fringe Area’ are shown as having equivalent densities (at the Australian Bureau of Statistics [ABS] Census Collector District level) despite geographical contexts with different accessibility and socio-economic characteristics. The figure also shows how density representations can be used, in this case, by an NSW government department, to create misleading understandings. While five density categories are presented, those areas containing over 25 dwellings per hectare are lumped into one broad category. Yet this category contains a variety of high and medium density built form types (terrace and semidetached housing, walk-up apartments, and highrise apartments). Meanwhile, four categories represent essentially just one built form type: detached dwellings. The illusion is a Sydney comprising a heterogeneous mix of urban forms. Despite the obvious inaccuracies that come from using density calculations to represent differences in urban structure and form, as we have noted, density measures continue to be used because they enable researchers to employ with relative ease the statistical tools which assess the level of association between two or more sets of numeric variables. Although it could be argued that calculating linear regression is the statistically correct approach to © 2014 Institute of Australian Geographers

© 2014 Institute of Australian Geographers

Figure 1

Comparison of area with same density – Sydney metropolitan area (adapted from NSW Department of Planning, 2005).

FRINGE AREA 9.01-12.00 dwellings per hectare

INNER AREA 9.01-12.00 dwellings per hectare


445

446 showing correlation and suspected relationship, pre-existing research shows there are many potential influences on sustainable outcomes, only one of which is urban structure and form. Moreover, as we well know, it is extremely difficult to show causality between correlated variables, perhaps more so in urban research. Indeed, attempting to prove causality between, say, high density and low energy use, when so many other behavioural influences are at play, could well be futile. Fortunately, there are statistical tools available to establish statistical association without implying causality, depending on the type of variables being investigated. These encourage more textured accounts of urban life, including consideration with other possible influences on sustainability outcomes, including levels of education, beliefs and attitudes, and a range of socio-economic variables. Importantly, such a willingness to embrace complexity in urban research also sends a clear message to planners and policy makers that achieving sustainable urban outcomes involves comprehensive approaches beyond simplistic density-raising actions. Here we review efforts to better understand the complexities of urban sustainability. We find an increasing interest in developing more empirically based urban structure and form typologies, incorporating variables such as urban structure, residential type, public transport accessibility, and land use mix (Newton, 2000; Holden and Norland, 2005; Jenks and Jones, 2010) as variables. Ghosh and Vale (2009), for example, develop a taxonomy for Auckland, New Zealand, that uses five scales to describe urban mix in large metropolitan areas (see Figure 2): metropolitan/regional scale; sub-metropolitan/ city scale; community/neighbourhood scale; local/residential block scale; and house/micro scale. When applied to Sydney, Ghosh and Vale’s taxonomy usefully distinguishes an area on the fringe, with little public transport access from an area located adjacent to a rail corridor with good public transport access, even though both areas might have the same residential density, as can be observed in Figure 1. Obviously, differences in transport modal choices for residents living in these two situations would produce significant sustainability outcomes. Indeed, Owens (2005) shows this to be indeed the case by documenting hundreds of neighbourhoods across the US that experience enormous variation in environmental quality, but with similar density and land use characters.


Applying an urban structure and form typology to Sydney, Australia Urban structure is the manner in which land uses are distributed throughout a city and urban form is the characteristic morphology of settlement of a city. The former could be considered the macro structural context within which micro examples of urban form operate. Significantly, there are some researchers who use density measures as an indicator of urban structure and form and who recognise that there are issues in respect to the representativeness of density calculations across large metropolitan areas. Another measure many researchers have used to represent urban structure, especially those seeking to qualify the relationship between density and transport, is distance to Central Business District (CBD). The interest among this group is in the influence of urban structure on transport choice (Mees, 2009, 11; Rickwood and Glazebrook, 2009, 184). The attraction of the distance to CBD variable is that it enables the researcher to continue to calculate lineal relationships. Yet this approach compounds the original problem by validating the density variable rather than seeking alternatives to its use. As well, in many large cities like Sydney, the distance to CBD is problematic because it does not capture the urban structure differences across a polycentric metropolitan area. Sydney contains many centres of varying size located along its radials – including larger employment centres like Parramatta and Chatswood, and sub-regional centres like Bankstown and Liverpool. This variety and disjuncture encourages an alternative approach to depicting urban structure and form, and our attempt is described in this section. We take a lead from Newton (2000), one of the few Australian researchers to apply an urban structure typology to a large metropolitan area (Melbourne); in his case, for examining the energy and greenhouse gas emission implications of different urban structures. Newton nominated six different urban structure types, based on ‘archetypal urban geometries’identified by others (including Pressman, 1995 and Minnery, 1992) to describe current and possible future urban growth: the dispersed city, the compact city, the multi-node city, the corridor city, the fringe city, and the ultra city (Newton, 2000, 47; see Table 1). These are similar to the categories nominated by Ghosh et al. (2006) in their New Zealandbased typology for the sub-metropolitan/city scale (see Figure 1). Although Newton uses density to describe each category, it is not the only © 2014 Institute of Australian Geographers


Figure 2

447

Ghosh and Vale’s (2009, 514) urban structure form/taxonomy scale.

consideration – location relative to the CBD and accessibility are also important. For Melbourne, Newton was interested in the ways different urban structure categorisations © 2014 Institute of Australian Geographers

affect sustainability outcomes (Newton, 2000). After testing each urban structure type in relation to indicators such as energy consumption and vehicle kilometres travelled, Newton concluded

448 Table 1


Newton’s urban structure categories.

Urban Structure Category

Newton’s Description (Melbourne Context)

Compact city Multi-node city

Increased population and density in the inner group of suburbs surrounding CBD. Increased population, housing densities and employment at selected nodes across the metro area, with key infrastructure linking the nodes. Growth along linear corridors emanating from the CBD and supported by upgraded transport infrastructure Outward random suburban expansion at relatively low densities, with connections to the central cities as the key economic node Additional growth beyond the dispersed suburban context Additional growth accommodated primarily in provincial cities within 100 km of the principal city and linked by high-speed rail

Corridor city Dispersed city Fringe city Ultra city

Source: Newton, 2000, 47.

that urban structure does matter, such that a trend to a more compact city, however defined, will lead to significant environmental improvements, compared with the business-as-usual dispersed city model (Australian Academy of Technological Sciences and Engineering, 1997, 170; Newton, 2000). Our research uses the ABS geographical classification for Urban Centre/Locality to define the Sydney metropolitan area (see Figures 3 and 4). It is apparent when viewing Sydney’s historical geography that each of Newton’s urban structure categories can capture different parts of the metropolitan area. The state government’s Metropolitan Plan for Sydney 2036 (NSW Department of Planning, 2005) contains a map showing Sydney’s urban growth history which broadly reflects the type of urban structure categories Newton describes (reproduced in Figure 3). The suburbs shown in dark red are areas developed before 1917 and possess the inner-area compact city, multi-node, and corridor city characteristics described by Newton. The areas contain much of Sydney’s older apartment, semi-detached, terrace and mixed use land use stock, as well as newer high-rise apartment buildings, especially in and around the CBD and in the multi-node locations. The compact city, multi-node, and corridor city areas also have excellent levels of walking and public transport accessibility. Areas developed between 1917 and 1945 are shown in lighter red and represent the dispersed city category, areas where suburbs dominated by detached dwellings first started to appear beyond Sydney’s rail corridors. The lighter pink areas represent the fringe city categories, which contain recently erected large-sized detached dwellings well away from any railway lines. New master-planned residential estates dominate

these areas. The yellow areas are areas designated for future fringe growth. Although Sydney’s historical urban growth rings are a reasonable guide for assigning Newton’s urban structure categories, we used Geographical Information Systems (GIS) for a more detailed examination of potential urban structure category boundaries in an attempt to place every Sydney Census Collection District (CCD) within its most appropriate urban structure category. Figure 4 and Table 2 specify the locations of each urban structure category, using local government area boundaries, public transport walking catchments, major road corridors, and natural area boundaries. In this exercise, it becomes clear that an additional urban structure category is required. A ‘secondary centre’ urban structure category has been added to reflect Sydney’s polycentric structure. This structure emerges from the ‘city of cities’ approach of the Metropolitan Strategy (NSW Department of Planning, 2005), which has resulted in the strengthening of subregional employment and retail centres located on rail corridors and major bus corridors in the Sydney metropolitan region. Sydney’s secondary centres – namely, Burwood, Ashfield, Hurstville, Bankstown, Penrith, Kogarah, Blacktown, Dee Why, and Sutherland – are smaller than the larger multi-node regional centres which have been established for some time (Parramatta, Chatswood, Hornsby, Liverpool, and Campbelltown), but larger than the village-like centres located within the corridor city structure. In order to better understand the level of urban form differentiation that exists within each urban structure category at the level of the residential block/local scale (as per Ghosh and Vale’s © 2014 Institute of Australian Geographers

Figure 3


1917 - 1945

1945 - 1975

1975 - 2005

Hurstville

Sydney’s urban growth history. Source: 2005 city of cities metro strategy, NSW Department of Planning, 2005.

BEFORE 1917

Parramaa

Epping

2031 – if current rate of urban growth connues

Chatswood


449

Sydney’s urban structure locations. Map source: UWS Urban Research Centre, 2012.


Figure 4

450



Table 2 region.

Urban structure locations in Sydney metropolitan

Category

Where Applied

Compact

CCDs located within Sydney’s inner urban LGAs – City of Sydney, Leichhardt, Waverley, Marrickville, and Randwick (north of Gardener’s Road). These LGAs are generally highly accessible to public transport, employment, retail and essential services, with a mixture of housing choice. MultiCCDs located within 800 m of railway node stations (proxy for centroid) located within designated (by Planning NSW) high-order centres: Chatswood, Parramatta, Hornsby, Liverpool and Campbelltown. Similarly accessible to public transport, employment, retail and essential services as the Compact City but dominated by four-storey and above apartments. SubCCDs located within 800 m of railway regional stations (proxy for centroid) located within designated (by Planning NSW, 2005) secondary centres: Bankstown, Ashfield, Burwood, Blacktown, Penrith, Sutherland, Dee Why, Kogarah and Hurstville. In the case of Dee Why, it is all CCDs within 800 m of Pittwater Road and Dee Why Parade intersection. Sub-regional centres have similar characteristics to the multi-node city, but on a smaller scale, with less employment opportunities. Corridor Any CCDs outside the previous categories that are located within 800 m walking distance of a railway station or 400 m of a high-order bus corridor (Anzac Parade and Pittwater Road). Characterised by good accessibility to local shopping precincts with a mixture of housing choice (other than four-storey and above apartments). Dispersed Areas constituting Sydney’s traditional suburban environment (see Figure 3– suburban areas developed between 1917 and 1976). Generally comprises a mixture of old and contemporary detached dwellings dominated by car access and local bus services. Fringe Designated areas beyond the traditional dispersed suburban environment. Primarily consists of contemporary build, with poor public transport and local service accessibility. Reference: NSW Department of Planning, 2005. CCD = Census Collection District; LGA = Local Government Area. © 2014 Institute of Australian Geographers

451

typology shown in Figure 1), we then examined dwelling type characteristics. We grouped housing types listed by the ABS into four categories: detached dwellings; semi-detached or terrace style dwellings; up to three-storey walk-up apartments; and four-storey and above apartments. Table 3 shows that there are notable differences in the proportions of these dwelling types contained within each urban structure category. These differences include detached dwellings predominating in the fringe (86% of all dwellings counted) and dispersed categories (66% of all dwellings counted), and walk-up and four-storey and above apartments predominating in the multi-node (64% of all dwellings counted) and sub-regional centre (72% of all dwellings counted) categories. The methodology adopted for this research project required secondary data to be attained to begin testing levels of association that may exist between the nominated urban structure and form categories and data representing consumption behaviours that affect the environment. Three indicators were chosen to test potential associations with urban structure and form: electricity consumption per person, water consumption per person, and car ownership per household. We were able to attain AusGrid’s (formerly Energy Australia) 2006 aggregate household electricity consumption data; Sydney Water’s 2006 aggregate household water consumption data; and ABS’s 2006 car ownership data. The use of these data is highly novel and represents a major advance in the integration of actual population data at a fine scale into an investigation of urban sustainability questions. These three consumption data categories (electricity, water, and transport) also correspond to behaviours identified as the critical contributors to household-related greenhouse gas emissions (Holden and Norland, 2005). Each secondary data set was aggregated to the ABS Census CCD level. As such, inference about the influence of dwelling type on the sustainability data is only valid when only one dwelling type represents the whole of the CCD. In order to analyse the combination of urban structure and dwelling type (representing urban form) for our research project, we isolated those CCDs that contain overwhelmingly only one housing type. As a result, we were unable to compare different dwelling types within every CCD located in the Sydney region. Rather, our analysis is confined to those CCDs within different urban structure categories that contain a single dwelling/urban form type. Table 4 shows

452 Table 3


Sydney’s urban structure categories – number of dwellings and percentage of housing type. Detached

Category Compact Multi-node Sub-regional Corridor Dispersed Fringe

Number 40 892 4079 2773 148 514 168 703 45 884

% 22 30 20 52 66 86

Semi-Terrace

Walk-Up Apartments

Number 40 771 675 1046 32 020 28 357 4057

Number 47 707 3708 6897 70 794 43 237 2433

% 22 5 8 11 11 8

% 26 27 50 25 17 5

Four-Storey Plus Apartments

Other

Number 50 651 5086 3119 34 361 15 969 530

Number 1968 20 61 1843 760 180

% 28 37 22 12 6 1

Total

% 1 0 0 0 0 0

Number 181 989 13 568 13 896 287 532 257 504 53 084

% 100 100 100 100 100 100

Source: ABS Census, 2006.

Table 4

Sydney’s urban structure categories – number of predominant housing type CCDs (above 95%).

Compact Multi-node Subregional Corridor Dispersed Fringe Total

Detached

Semi-Terraces

Walk-Up Apartments

Four-Storey Plus Apartments

Total

13 8 1 303 581 531 1429

9 0 0 4 10 6 29

11 9 4 24 21 0 68

71 13 3 19 4 0 110

104 30 8 350 616 537 1645

Source: ABS Census, 2006.

that we were able to locate sufficient numbers of CCDs with one dominant housing type (where a CCD contains above 95% of one particular dwelling type) to enable dwelling type to be analysed and compared across the metropolitan region. Statistical tests using a typology approach As we indicate above, we question the need for establishing neat lineal relationships between density and different sustainability indicators in urban form research. Rickwood (2009, 64) states that the complexity of urban systems often does not easily enable theoretical models to be developed, and, as a consequence, there is resort to descriptive statistical and econometric approaches to make sense of available data. The problem, according to Rickwood, is that a reliance on easily computable metropolitan-scale variables (such as density calculations) can stymie an understanding of urban form differences. But even then, says Rickwood, the results from such studies can be informative. We reiterate our concern stated above, however, that privileging density in urban form studies, even if surrounded by caveats, sends the crude message to planners and policy makers that densification is critical in achieving improved sustainability outcomes.

Table 5

Choosing the correct statistical test. Dependent Variable Categorical Continuous

Independent Categorical Chi-Square t-Test, ANOVA variable Continuous LDA, QDA Regression Source: Ludford, 2012. LDA = linear discriminant analysis; QDA = quadratic discriminant analysis.

Descriptive variables such as the urban structure and form categories developed for this research project cannot be sequentially ordered or differentiated from each other using a mathematical method (Ludford, 2012). They are therefore categorical. On the other hand, the secondary sustainability data that we use to compare different urban structure and form categories are differentiated numerically, so for statistical purposes they are continuous. Table 5 shows that there are two tests available – t-tests and analysis of the variance of means (ANOVA) – for calculating the level of association between our categorical variable (urban structure and form categories) and the variety of © 2014 Institute of Australian Geographers

453


Mean Electricity Consumpon in Urban Structure Categories (yearly kW per person)

Fringe Dispersed Corridor Sub-regional Mul-node Compact 0

Figure 5

500

1000

1500

2000

2500

3000

3500

4000

2006 electricity consumption per person by urban structure. Data source: AusGrid, 2011.

Mean Water Consumpon in Urban Structure Categories (yearly kL per person)


Figure 6

74

76

78

80

82

84

86

Water consumption by urban structure. Source: Sydney Water, 2011.

continuous data available to represent sustainability behaviours. If our urban structure and form typology contained only two categorical independent variables (e.g. compact city and dispersed only), then a two-sample t-test would be the correct statistical test. However, because the categorical independent variable has more than two variables (there are six urban structure categories), ANOVA should be applied. Even though ANOVA does not assist in explaining the cause of difference that might exist across urban form categories, it is an approach that is appropriate in urban form research for assessing statistically significant patterns of difference (Handy, 1996; Sobhani, 2009). If statistically significant patterns of difference emerge across the urban structure and form categories and across the three secondary sustainability variables, then further research can take place to understand potential causes of such differentiation. © 2014 Institute of Australian Geographers

Our analysis reveals statistically significant associations between our sustainability data (electricity consumption, water consumption, and car ownership rates) and the urban structure categories (see Figures 5–7) For electricity use, one-way ANOVA tests showed that there are significant differences in the means between most urban structure categories, except between the compact city and multinode categories, the multi-node and corridor categories, and the multi-node and sub-regional categories. For water use, there is statistically significant difference in the means between compact city and every other category. For car ownership there is significant difference in means between most urban structure categories, except between the compact city and subregional and multi-node categories. On the surface, the urban structure variable appears to have an influence on the three key

454


Car Ownership rates in Urban Structure Categories (cars per household per person)


Figure 7

Table 6

0.5

1

1.5

2

Car ownership by urban structure. Source: ABS, 2006.

Two-way ANOVA testing for electricity, water, and car ownership considering urban structure and dwelling.

Dwelling Variable

Detached dwelling Semi or Terrace Walk-ups Four-storey above

Electricity

Water

Car Ownership

F Value

Sig.

F Value

Sig.

F Value

Sig.

17.8851 0.399 0.551 1.938

0.00 0.67 0.70 0.10

21.7221 0.138 2.8391 7.8271

0.00 0.94 0.02 0.00

33.7931 2.125 3.361 6.0341

0.00 0.10 0.01 0.00

1

Indicates statistical significance. Source: PASW SPSS Statistics 18. Each F tests the simple effects of urban form within each level combination of the other effects shown. These tests are based on the estimable linearly independent pairwise comparisons among the estimated marginal means.

sustainability indicators in different ways. Yet, can it be assumed that the urban structure category itself is responsible for generating statistically significant difference in the means for each of the indicators, or are there other factors contributing to the results? From the relevant literature, it is apparent that two potential influences on the three indicators are dwelling type (representing urban form difference in this research project) and socio-economic status (Randolph and Troy, 2007; 2008). With regard to dwelling type, the different heating and cooling requirements of different building structure types, in particular, can have a marked impact on electricity use. Socioeconomic factors are also believed to be associated with energy consumption, with Randolph and Troy (2007) showing for Sydney that lowdensity suburbs with higher incomes record much higher per capita electricity consumption than low-density suburbs with lower and moderate incomes. Water consumption rates are also

affected by dwelling type. Randolph and Troy (2008) demonstrate that those living in detached dwellings with backyards, with more opportunity for discretionary water use, have a greater capacity to lower consumption levels. With regard to car ownership, socio-economic factors and dwelling type may also have an influence on transport modal choice, although urban structure is identified as having a more significant influence on transport modal choice than these factors (Mees, 2009; Rickwood and Glazebrook, 2009). What we can do to understand the level of influence of one category (urban structure) over another (dwelling/urban form or Socio Economic Indexes for Areas [SEIFA]) is perform a two-way ANOVA test. Tables 6 and 7 show the results of two-way ANOVA tests when the effect of urban structure is assessed for both the dwelling/urban form variable and SEIFA index variable. Table 6 shows that urban structure has a statistically significant influence on mean electricity consumption in detached dwellings, while mean © 2014 Institute of Australian Geographers

455


Table 7

Two-way ANOVA testing for electricity, water, and car ownership considering urban structure and SEIFA index.

SEIFA Variable

Ranking 1 Ranking 2 Ranking 3 Ranking 4 Ranking 5 Ranking 6 Ranking 7 Ranking 8 Ranking 9 Ranking 10

Electricity

Water

Car ownership

F Value

Sig.

F Value

Sig.

F Value

Sig.

0.39 1.52 21.5321 19.5351 0.50 1.39 4.3281 3.8151 13.6111 48.4361

0.68 0.21 0.00 0.00 0.73 0.24 0.00 0.00 0.00 0.00

21.5261 2.334 1.46 1.171 1.826 3.9021 4.4721 6.1921 3.971 9.3551

0.00 0.04 0.20 0.32 0.10 0.00 0.00 0.00 0.00 0.00

32.2411 13.9361 17.1521 28.6061 45.071 69.2271 95.8721 112.1031 221.6061 497.8771

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

1

Indicates statistical significance. Source: PASW SPSS Statistics 18. Each F tests the simple effects of urban form within each level combination of the other effects shown. These tests are based on the estimable linearly independent pairwise comparisons among the estimated marginal means. SEIFA = Socio Economic Indexes for Areas.

electricity consumption does not vary significantly for the other dwelling types across different urban structure contexts. In other words, electricity consumption has the potential to vary significantly in detached dwellings when placed in different urban structure contexts. The table also shows that urban structure has a significant influence on water consumption in three of the four dwelling types, with the semi-detached and terrace dwelling type being the only exception. As well, urban structure influences car ownership for all dwelling types, except semi-detached and terraces. Table 7 also shows that urban structure has a statistically significant influence on electricity usage and water usage across many of the SEIFA index ranking categories, and for all SEIFA rankings when it comes to car ownership. In fact, Table 7 shows that for every SEIFA ranking above 7, urban structure has a statistically significant influence over electricity consumption, water consumption, and car ownership. These results led us to choose case study locations around Sydney representing different urban structure and form (dwelling) types so as we could obtain primary data from households to interrogate why such statistically significant associations might exist. Self-completion questionnaires were delivered to every household in the case study areas. The questionnaire contained mostly multiple choice questions to enable easy comparison of answers across case-study areas. The questions were developed with reference to © 2014 Institute of Australian Geographers

pre-existing large surveys such as the 2006 and 2011 ABS Census, Newspoll surveys on climate change (2011); the GreenLight Survey (Sustainability Victoria, 2006); and the NSW Department of Environment’s ‘Who Cares About the Environment Survey’ (2006 and 2009). An overall response rate of 23% was achieved, which is considered satisfactory, considering the good level of representativeness we attained when comparing population profile data of case study respondents with known ABS census population profile data (such as tenure type, language spoken at home, and level of education). The survey of case study areas generated responses regarding behavioural choices that are also defined as categorical for statistical purposes (i.e. they are not continuous or ordinal). Whereas t-tests or ANOVA are most appropriate when determining the level of association between a categorical variable (urban structure and form) and continuous or ordinal data (the sustainability data), chi-square tests should be used when testing two categorical variables (see Table 5). The categorical responses from our questionnaire survey were divided into dependent or independent variables in order to develop crosstabulations. The chi-square cross-tabulations then enabled us to quantify the most statistically significant associations between urban structure and form, and the variety of sustainable behaviour data we were able to collect. The chisquare testing showed that urban structure and form have a strong association with transport

456 behaviours, especially when compared with other potential influences like attitudes and beliefs and a range of socio-economic factors. On the other hand, urban structure and form are two of several factors (including attitudes and beliefs) that are statistically associated with reduced electricity and water consumption. The chi-square results suggest that the case study locations representing different urban structure and form categories may be statistically associated with other independent variables such as belief in human-induced climate change, level of education, and a variety of socio-economic factors, which manifest as geo-political differences across large metropolitan areas. This, then, might explain differences in detached dwelling electricity consumption across the urban structure categories. The nexus between urban structure and form and the combination of political values, level of education, and socio-economic factors, is an area of further research worth exploring. Conclusion This paper has assessed two different approaches to representing urban structure and form difference: density calculations and a typology approach. Given the inherent inaccuracies involved in using density calculations to represent different urban form types when assessing their relationship with sustainability data, it was decided to develop and test a typology approach to more accurately represent urban structure and form difference. In developing an urban structure and form typology for our case study location, the Sydney metropolitan area, we relied on Newton’s (2000) urban structure categories and adapted these to the Sydney context. We developed six distinct urban structure categories that we found existed in various parts of the Sydney metropolitan area and mapped these as accurately as possible with the assistance of GIS. We then placed every CCD within the Sydney metropolitan area into an appropriate urban structure category and highlighted those CCDs with a dominant housing type to enable a comparison of different combinations of urban structure and dwelling type (urban form). Applying an alternative descriptive typology for representing urban structure and form difference did not permit us to infer causal relationships between our urban structure and form categories and the sustainability data we used. We argue, however, that employing alternative


statistical approaches that calculate relative levels of association between categorical urban structure and form variables and a variety of sustainability data is more appropriate within an urban research context. When we undertake twoway ANOVA testing to ascertain the extent to which the urban form (represented by different dwelling type) and the SEIFA index (representing socio-economic differences) variables affect urban structure’s influence over the three sets of data, it becomes evident that urban structure, as a separate statistical variable, does influence household behaviours, especially car ownership, regardless of dwelling type (urban form) or socio-economic ranking. However, further case study results suggest that other variables such as belief in human-induced climate change, level of education, and a variety of socio-economic factors may manifest as geopolitical differences across large metropolitan areas and explain the level of association the urban structure variable has with electricity and water consumption. Such understandings would not be possible by using density calculations alone to represent urban structure and form difference. Our work demonstrates that the relationship between urban form and sustainability is not a simple one. Certainly, as we stress, the relationship cannot be reduced to a single density metric. Large urban areas are complex, historical places containing enormous variance in the sustainability behaviours of residents and households. Our research shows that we can find close relationships between these behaviours and certain types of urban form and structure. But we also find that substantial variation in sustainability outcomes are explainable to varying degrees by personal and household characteristics, independent of urban form and structure. To be clear, we are not saying that density does not matter. Rather, we are saying that it matters in certain circumstances, and policy measures to improve urban sustainability outcomes should be mindful of when and how things other than density, right down to personal attitudes and ideologies, become important. ACKNOWLEDGEMENTS The electricity and water consumption data used for this research was provided by AusGrid and Sydney Water respectively. This research would not have been possible without their assistance. REFERENCES ABS. 2006: Census of Population and Housing. Canberra.



Anderson, W., Kanaroglou, P. and Miller, E., 1996: Urban form, energy and the environment: a review of issues, evidence and policy. Urban Studies 33, 7–35. Ausgrid. 2011: 2006 Aggregate Electricity Consumption of households in Sydney catchment area. Ausgrid, Sydney. Australian Academy of Technological Sciences and Engineering, 1997: Urban Air Pollution in Australia. Australia Commonwealth Government, Canberra. Bamford, G., 2007: Understanding Housing Density. 2007 Housing Conference, University of Queensland, Brisbane. Bamford, G., 2009: Urban form and housing density, Australian cities and European models: Copenhagen and Stockholm reconsidered. Urban Policy and Research 27, 337–356. Bramley, G., Dempsey, N., Power, S., Brown, C. and Watkins, D., 2009: Social sustainability and urban form: Evidence from five British cities. Environment and Planning A 41, 2125–2142. Brundtland Commission, 1987: Our Common Future: Brundtland Report. United Nations, Oxford. Burgess, R., 2000: The compact city debate: a global perspective. In Jenks, M. and Burgess, R. (eds) Compact Cities: Sustainable Urban Forms for Developing Countries. Spon Press, London, 9–24. Churchman, A., 1999: Disentangling the concept of density. Journal of Planning Literature 13, 389–411. Ewing, R. and Fong, R., 2010: The impact of urban form on U.S. residential energy use. Housing Policy Debate 19, 1–30. Forster, C., 2006: The challenge of change: Australian cities and urban planning in the new millennium. Geographical Research 44, 173–182. Florida, R., 2012: America’s truly densest metros [Online]. The Atlantic Media Company. Retrieved: 25 October 2012 from . Ghosh, S. and Vale, R., 2009: Typologies and basic descriptors of New Zealand residential urban forms. Journal of Urban Design 14, 507–536. Ghosh, S., Vale, R. and Vale, B., 2006: Domestic energy sustainability of different urban residential patterns: a New Zealand approach. International Journal of Sustainable Development 9, 16–37. Gleeson, B., Darbas, T. and Lawson, S., 2004: Governance, sustainability and recent Australian metropolitan strategies: a socio-theoretic analysis. Urban Policy and Research 22, 345–366. Gray, R., Gleeson, B. and Burke, M., 2008: Urban consolidation and household greenhouse emissions: towards a full consumption impacts approach. Urban Research Program Working Papers, Griffith University. Griffiths, D., 2009: Density targets: measuring everything except that which makes life worthwhile? International Cities Town Centres and Communities Society 2009, Geelong. Hall, P., 1997: The future of the metropolis and its form. Regional Studies 31, 211–220. Handy, S., 1996: Understanding the link between urban form and nonwork travel behavior. Journal of Planning Education and Research 15, 183–198. Holden, E. and Norland, I., 2005: Three challenges for the compact city as a sustainable urban form: household consumption of energy and transport in eight residential areas in the Greater Oslo Region. Urban Studies 42, 2145– 2166.


457

Jenks, M. and Jones, C., 2010: Dimensions of the Sustainable City. Future City, Vol. 2. Springer, New York, NY. Jones, C., Jenks, M. and Bramley, G., 2010: Complementaries and Contradictions. In: Jenks, M. and Jones, C. (eds) Dimensions of the Sustainable City. Future City 2, Springer, New York, NY, 239–256. Ludford, P., 2012: An Overview: Choosing the Correct Statistical Test. Department of Computer Science and Engineering, University of Minnesota, Minneapolis. Retrieved: 18 September 2012 from . Mees, P., 2009: How dense are we? State of Australian Cities Conference, Perth. McLoughlin, J. B., 1991: Urban consolidation and urban sprawl: a question of density? Urban Policy and Research 9, 148–156. Minnery, J., 1992: Urban form and development strategies: equity, environmental and economic implications. National Housing Strategy. Background Paper 7 ed, Canberra. Neuman, M., 2005: The compact city fallacy. Journal of Planning Education and Research 25, 11–26. Newman, P. and Kenworthy, J., 1989: Cities and Automobile Dependence: A Sourcebook. Gower, Brookfield, VT. Newman, P. and Kenworthy, J., 1999: Sustainability and Cities: Overcoming Automobile Dependence. Island Press, Washington, DC. Newman, P. and Kenworthy, J., 2000: Sustainable urban form: the big picture. In: Williams, K., Burton, E. and Jenks, M. (eds) Achieving Sustainable Urban Form. E and F N Spon, London, 109–120. Newton, P., 2000: Urban form and environmental performance. In Katie Williams, E.B. and Jenks, M. (eds) Achieving Sustainable Urban Form. E and F N Spon, London, 74–84. Newton, P., (ed.) 2011: Consumption and environmental sustainability. In Urban Consumption. CSIRO Publishing, Canberra, 173–197. NSW Department of Planning, 2005: City of Cities: A Plan For Sydney’s Future. NSW Department of Planning, Sydney. Retrieved: 14 April 2010 from . Owens, P., 2005: Beyond Density: Measuring Neighborhood Form in New England’s Upper Connecticut River Valley. PhD, University of California, Berkeley, USA. Perkins, A., Hamnett, S., Pullen, S., Zito, R. and Trebilcock, D., 2009: Transport, housing and urban form: the life cycle energy consumption and emissions of city centre apartments compared with suburban dwellings. Urban Policy and Research 27, 377–396. Pressman, N., 1995: Forces for spatial change. In Brotchie, J., Newton, P., Hall, P. and Nijkamp, P. (eds) The Future of Urban Form – The Impact of New Technology. Croom Helm Publishing, London, 349–361. Randolph, B., 2004: The changing Australian city: new patterns, new policies and new research needs. Urban Policy and Research 22, 481–493. Randolph, B., 2006: Delivering the Compact City in Australia: current trends and future implications. Urban Policy and Research 24, 473–490. Randolph, B. and Troy, P., 2007: Energy consumption and the built environment: a social and behavioural analysis. City Futures Research Centre Working Paper, University of NSW, Sydney. Randolph, B. and Troy, P., 2008: Attitudes to conservation and water consumption. Environmental Science and Policy 11, 441–455.

458 Randolph, B., Holloway, D., Pullen, S. and Troy, P., 2007: The Environmental Impacts of Residential Development: Case Studies of 12 Estates in Sydney. City Futures Research Centre, University of New South Wales, Sydney. Rickwood, P., 2009: Residential operational energy use. Urban Policy and Research 27, 137–155. Rickwood, P. and Glazebrook, G., 2009: Urban structure and commuting in Australian cities. Urban Policy and Research 27, 171–188. Sobhani, L., 2009: Paramatric and nonparametric statistical procedures of ANOVA in urban studies. Geography and Environmental Studies 1, 70–112.


Sustainability Victoria. 2006: 2006 GreenLight Report. Melbourne: Sustainability Victoria, Melbourne. Sydney Water. 2011: 2006 Aggregate Water Consumption of households in Sydney Catchment Area. Sydney Water, Sydney. Talen, E., 2011: Sprawl retrofit: sustainable urban form in unsustainable places. Environment and Planning B: Planning and Design 38, 952–978. Troy, P., 2004: Structure and Form of the Australian City: Prospects for Improved Urban Planning. Urban Policy Program. Griffith University, Brisbane, QLD.
