Mixed Land Uses White Paper

download Mixed Land Uses White Paper

of 39

Transcript of Mixed Land Uses White Paper

  • 8/11/2019 Mixed Land Uses White Paper

    1/39

    Carolina Transportation Program

    White Paper Series

    The Measurement of the Level of Mixed Land Uses:

    A Synthetic Approach

    Yan Song* and Daniel A. Rodrguez**

    Department of City and Regional Planning

    New East Bldg, CB#3140

    University of North Carolina

    Chapel Hill, NC 27599-3140; USA

    * Email: [email protected]

    ** Email: [email protected]

    Preparation of this White Paper was supported by a grant from the Robert WoodJohnson Foundation Active Living Research program.

  • 8/11/2019 Mixed Land Uses White Paper

    2/39

    1

    The Measurement of the Level of Mixed Land Uses:

    A Synthetic Approach

    Abstract

    Despite the burgeoning interests in studying mixed land uses and their

    relationship with individual and community outcomes in disciplines such as

    landscape ecology and the environment, transportation, health outcomes, and

    housing markets, there is a paucity of research on the measurement of such

    mixed land use. In this paper we provided a synthetic examination of an array

    of land use mix measures which would tap various dimensions of the urban

    land use mixture. We classified existing indices as measures of accessibility,

    intensity and pattern. With the purpose of evaluating the measures, we also

    applied selected measures in an empirical case study. Our review and the

    empirical application provide insights for researchers and practitioners

    regarding the appropriateness of particular measures for particular purposes.

    We propose three criteria for choosing the measures: the extent to which a

    measure captures the presence or configuration of land uses, practical

    considerations including data collection, amount of computation and ease of

    communicability, and connection between the measures and the purpose of the

    investigation.

  • 8/11/2019 Mixed Land Uses White Paper

    3/39

    2

    1. Introduction

    The separation of land uses has been the cornerstone of conventional land use

    planning in the U.S. Partly as a response to a set of complex problems brought

    on by urban sprawl that have beset most U.S. metropolitan areas, planners and

    researchers have begun advocating for the mixing of certain types of land uses.

    For example, the Smart Growth Network, established under the auspices of the

    U.S. Environmental Protection Agency, promotes the mixing of residential and

    commercial uses as one of the ten principles of Smart Growth. The Congress

    of New Urbanism (CNU) also calls for: Neighbourhoods [to] contain a mix of

    shops, offices, apartments, and homes; land uses are mixed-use within

    neighbourhoods, within blocks, and within buildings (CNU, 2002). In

    addition, the US Centres for Disease Control and Prevention has identified

    mixing land uses as a strategy to promote active community environments

    (Centres for Disease Control and Prevention, 2005).

    The interest in mixing certain land uses stems from emerging empirical

    evidence suggesting that greater mixture of complementary land use types,

    which may include housing, retail, offices, commercial services, industrial and

    civic uses, is related to peoples propensity to walk and thus to be physically

    active, transit use, and property values. Mixed land uses also have been

  • 8/11/2019 Mixed Land Uses White Paper

    4/39

    3

    associated to lower automobile ownership, use and emissions. Although not

    tested empirically, mixed land uses also are thought to reinforce streets as

    public spaces, create a sense of community and local investment, assist in

    achieving local housing and employment mixes, and promote transit-

    supportive development among others (American Planning Association, 1998).

    Despite the practical interest and the mushrooming empirical research, there

    have been few substantive analyses devoted to the measurement of land use

    mixtures. In this paper we: a) provide a synthetic examination of land use mix

    measures used in prior research; and b) we adapt and test related measures

    used in other disciplines (ecology, sociology, business, micro-economics). By

    providing insights regarding the strengths and weaknesses of various land use

    measures, we contribute to clarifying existing evidence and provide

    suggestions for future researchers. Four sections follow in this paper. In the

    next section we summarize recent research on land use mixtures and outcomes

    of interest to planners and policy-makers. The second section presents our

    approach to categorizing, developing and implementing land use mix measures

    and discusses the strengths and weaknesses of the measures. In the third

    section we summarize an empirical application of various measures to a case

    study in Hillsboro, Portland metropolitan area (OR). We use the empirical

  • 8/11/2019 Mixed Land Uses White Paper

    5/39

  • 8/11/2019 Mixed Land Uses White Paper

    6/39

    5

    outcomes (physical activity behaviour, obesity), and housing markets (property

    values).

    Aided by the increasing abundance of micro-level data that provide a rich

    empirical basis, the relationship between land use mix and transportation

    outcomes has received a flurry of attention over the last decade. Having higher

    mixes of land uses nearby has been positively related to frequency of trips by

    pedestrian and bicycle modes (Cervero, 1996; Greenwald and Boarnet, 2001;

    Handy, 1996; Khattak and Rodriguez, In press; Kitamura et al., 1997) and

    negatively related to frequency of auto trips (Cervero and Kockelman, 1997).

    Discrete choice models of travel mode also have shown that high levels of land

    use mixing in ones home or work neighbourhood are related to higher

    walking, bicycling and transit shares (Cervero, 1996; Srinivasan, 2002),

    although the effect size has been qualified as fairly marginal (Cervero and

    Kockelman, 1997) and modest (Cervero and Duncan, 2003). Shorter

    commuting distances (Cervero, 1996) and lower commuting times (Ewing et

    al., 2003) have been positive related to mixed land uses. Finally, evidence of

    associations between mixed land uses and auto ownership is less consistent,

    with Ewing et al (2003) finding no relationship and others finding a negative

    relationship (Cervero, 1996; Hess and Ong, 2002).

  • 8/11/2019 Mixed Land Uses White Paper

    7/39

    6

    In landscape ecology, land uses are often the starting point of modelling

    approaches (ONeill et al., 1988; Turner, 1990). Because land uses are

    intimately associated with ecological consequences, there is an interest in

    quantifying land uses and potential changes. Furthermore, environmental

    consequences often vary depending on the pattern of uses, the remaining

    habitat, and the size and proximity of disturbances to sensitive areas

    (Geoghegan et al 1997). Thus, quantifying uses relative to each other, their

    pattern, is essential for monitoring and assessing ecological outcomes. In this

    vein, studies have attempted to examine the relationship between land use mix,

    emissions and air quality. Although some studies have found a positive

    relationship between mixed land uses and emissions (Frank et al., 2000), others

    have detected an opposite, negative relationship (Ewing et al., 2003).

    For health-related disciplines, the emergence of ecologic models (Stokols,

    1992) has underscored the levels at which multiple factors (personal,

    interpersonal, community, environment and policy) can influence individual

    behaviour and health outcomes. As a result, an expanded set of factors, such

    as neighbourhood land use mix, are hypothesized to influence individual

    behaviour (Sallis et al., 1997). Although land use mixing has been positively

  • 8/11/2019 Mixed Land Uses White Paper

    8/39

    7

    associated with physical activity time (Frank et al., 2005; Hoehner et al.,

    2005), the emerging evidence with respect to obesity is equivocal, with studies

    finding conflicting associations (Frank et al., 2004; Rutt and Coleman,

    Forthcoming). By contrast, the evidence regarding the relationship between

    physical activity and the mixing of residential and recreational land uses (like

    parks and community centres) more consistently shows a positive association

    (Giles-Corti et al., 2005; Giles-Corti and Donovan, 2002a; Giles-Corti and

    Donovan, 2002b).

    Land use mixes also have been related to housing markets and individuals

    preferences for housing types. Measures of land-use mixbetween residential

    and commercial uses generally correlate with high residential land prices

    (Cervero and Duncan, 2004; Geoghegan et al., 1997; Song and Knaap, 2004)

    and in related studies land prices and the mix between residences and open

    space are also positively related (Geoghegan, 2002; Irwin, 2002; Irwin and

    Bockstael, 2001).

    In summary, empirical ambiguities remain regarding the relationship between

    land use mixtures and community and individual outcomes. The presence of

  • 8/11/2019 Mixed Land Uses White Paper

    9/39

    8

    mixed commercial and residential uses of land appears to support non-

    motorized modes of travel, higher physical activity time, and higher property

    values. By contrast, the evidence regarding land use mix and auto ownership,

    obesity, and air quality is equivocal. Although these variations are likely due

    to differences in the context of each study, the type of behaviour being

    observed and the data used, differences in measurement, scale, and refinement

    of land use mixture also contribute to the distinct outcomes. It is thus

    necessary to scrutinize and evaluate various measures of land use mix used in

    various fields. In the next section we turn to summarizing existing measures of

    land use mix and adapting new measures that were developed in other fields,

    while discussing their strengths and weaknesses and suggesting potential

    refinements.

    3. Measures of land use mix

    Urban planners have developed numerous ways to study the level of land use

    mixture. Researchers from other fields have also developed loads of measures

    in studying the distributional characteristics of various phenomena. For

    example, economists have examined market share of firms; sociologists have

    observed residential segregation patterns, and landscape ecologists have

    monitored land covers in relations to each other. Many of these measures can

  • 8/11/2019 Mixed Land Uses White Paper

    10/39

    9

    be adjusted to serve our purpose of assessing land use mixture. For ease of

    summarizing, we categorize various measures based on their different

    approaches to conceptualizing land use mixture: to appraise mixture based on

    the concept of accessibility (or proximity), of intensity (or magnitude), and of

    distribution pattern. Accessibility is the degree to which mixed land activities

    are easy to reach by residents; intensity is the volume or magnitude of mixed

    land uses present in an area; and pattern is the way in which different types of

    land uses are organized in an area. Our discussion on measures of land use

    mix below revolves around these three concepts.

    For the purpose of demonstration, we divide land use into different types:

    single family residential (residential hereafter) and non-single family

    residential (non-residential hereafter). Non-residential land further includes:

    commercial stores, multi-family residential units, light industrial sites, public

    institutions, and public parks. The geographic units of analysis of

    measurement, depending on the measures, are either individual land parcel

    units, or neighbourhoods. In this study, neighbourhoods can be defined by

    census boundaries such as zip codes, census tract, blockgroups, or Traffic

  • 8/11/2019 Mixed Land Uses White Paper

    11/39

    10

    Analysis Zones (TAZs)1. Neighbourhoods can also be defined by user-

    determined boundaries such as individual buffers of user-defined sizes drawn

    around land parcels, or square grids of user-defined sizes. Figure 1

    demonstrates the organization chart of the measures included in this study. For

    each of the measures presented next, we include a detailed example of at least

    one (and in many cases more than one) implementation of the specific measure

    in the Appendix, including references to studies that have used such measures.

    --insert Figure 1 here--

    3.1. Accessibility-based land use mix measures

    3.1.1. Distance

    Definition:The linear or street network distance between residential land use

    and another given non-residential land uses.

    Unit of analysis:Individual land parcels or neighbourhoods.

    1A Traffic Analysis Zone (TAZ) is a special area delineated by state and/or local

    transportation officials for tabulating traffic related dataespecially journey-to-work and

    place-of-work statistics.

  • 8/11/2019 Mixed Land Uses White Paper

    12/39

    11

    This measure takes adjacent non-residential land uses into consideration by

    calculating the nearest distances between pairs of observations. It also

    accounts for individual variances in proximity to other land uses. However,

    the measure offers little information on the broader context of the proximity by

    paying no heed to land uses other than the closest one. In addition, the

    measure takes no notice of the size of the nearest non-residential land use.

    3.1.2. Gravity

    Definition:The simplest gravity-based measure of land use mix can be defined

    as the sum of accessibility of residential land use to all other given type of non-

    residential land uses, discounted by the distance decay function between these

    two points.

    Unit of analysis:Individual land parcels or neighbourhoods.

    This approach generates a relatively comprehensive measure of accessibility

    from a residential land use to a given type of non-residential land uses by

    including distances to all other non-residential units. A major challenge with

    this straightforward approach is to fine-tune the impedance function to reflect

    the true impedance at that point, since as urban structures change, the distance

    decay or impedance function also changes. Another limitation of this measure

    is that it overlooks the scale or the size of non-residential land use activities.

  • 8/11/2019 Mixed Land Uses White Paper

    13/39

    12

    3.1.3. Gravity with competition

    Definition:The sum of accessibility of residential land use to all other given

    type of non-residential land uses, discounted by the distance decay function

    between these two points, and extended by considering both the supply side of

    non-residential land uses (i.e., the attractiveness of the non-residential land

    use) and the demand side of non-residential land uses (i.e., the competition for

    consuming the functions provided by the non-residential land use).

    Unit of analysis:Individual residential units or neighbourhoods.

    This measure provides information on accessibility to non-residential land use

    in a more thorough way than the previous measures by considering both the

    scale (the attraction) of and the competition for the services. However, it

    assumes that accessibility is based only on the distance between various

    competitors and the destinations, and their relative attractiveness, for example

    as dictated by floorspace or number of employees.

    3.1.4. Denominator of destination choice model

    Definition:The denominator of a discrete model of destination choice can be

    interpreted as a generalized measure of accessibility to destinations.

  • 8/11/2019 Mixed Land Uses White Paper

    14/39

    13

    Unit of analysis:Individuals.

    This measure has theoretical appeal because it is rooted in consumer choice

    theory and thus can be linked directly to consumer surplus calculations of

    accessibility. The main drawback is that it requires substantial attribute data on

    all destinations or on a sample of likely destinations. To this end, data on the

    preferred destination and non-preferred (but available) destinations for a

    representative sample of individuals in the study area are necessary. Another

    limitation is that the comparability of this measure across samples or across

    individuals is limited because the utility function is not measured in a

    consistent scale.

    3.2. Intensity-based land use mix measures

    3.2.1. Counts

    Definition:Number of non-residential activities in the neighbourhood.

    Unit of analysis:Neighbourhoods.

    3.2.2. Area proportions

  • 8/11/2019 Mixed Land Uses White Paper

    15/39

    14

    Definition: Proportions of different types of land uses within a user-defined

    neighbourhood.

    Unit of analysis:Neighbourhoods.

    The measures in this and the previous category are easy to compute and offer

    practical information on the intensity of a particular type of land use in a user-

    defined neighbourhood. Unfortunately, there are several limitation of the

    analyses based at the neighbourhood level. First, as the counts or proportion of

    land uses are conventionally aggregated by areal units such as census

    boundaries and TAZs, fine variations at smaller-unit level are averaged out and

    smoothed over during successive levels of aggregation, effectively

    disappearing with each higher level of aggregation (the modifiable areal unit

    problem, MAUP).2

    2Researchers have attempted to manage MAUP by computing land use measures at the parcel

    level, identifying homogenous buffers around individual land parcels (e.g., residential housing

    units) as the parcels immediate neighbourhoods and thus avoiding the aggregation problem.

    Although this approach is well-founded in presenting a uniform comparison on land usemixture across the immediate neighbourhoods of land parcels, the appropriate size of the

    buffers remain in debate. If the purpose of examining land use mixture is to evaluate the

    availability of activities within walking distance of households, it is then generallyrecommended to use 1/4-mile as buffer radius since pedestrian access is generally accepted as

    1/4-mile network distance (Duany & Plater-Zyberk, 1992). One criticism of this uniform

    buffer-drawing approach is that it assumes people with different characteristics (e.g., teens vs.

    adults) and at different locations (e.g., urban core vs. exurban) perceive their neighbourhoodsto be of equivalent sizes. However, it is more likely that neighbourhood sizes deviate fromeach other within the urban landscape and between different population groups. It should be

    noted that there is a paucity of research quantifying the size of relevant catchment areas as

    immediate neighbourhoods, particularly for the purpose of studying behaviour.

  • 8/11/2019 Mixed Land Uses White Paper

    16/39

    15

    Second, it is well understood that results are determined by the oftentimes

    arbitrary location of neighbourhood boundaries and therefore might be

    misleading. It is also necessary to consider how different levels of aggregation

    can affect results. For example, a larger neighbourhood is simply prone to

    more land use types. If the results change with the selection of different sizes

    of areal units, the reliability of results is called into question.

    Third, there is concern with using larger neighbourhoods (e.g., census tracts) is

    that the units of analysis are too large to have an intrinsic meaning with respect

    to the underlying land use distribution. The issue the non-uniformity of

    space has to do with the fact that the physical environmental conditions need

    to be taken into account as contexts for confirm or refute calculated

    distribution patterns. For example, the observed concentration of residential

    and non-residential uses can be less significant than originally thought because

    the other part of the city has a large lake.

    3.3. Land-use mix pattern measures

  • 8/11/2019 Mixed Land Uses White Paper

    17/39

    16

    Compositional pattern of land use mixture, as manifested through spatial

    assimilation of land development, is another important aspect of studying

    mixed land uses. We now present the measures of pattern, which can be

    further classified into three dimensions: evenness, exposure, and clustering

    (Figure 1).

    3.3.1. Evenness and Diversity

    Evenness and diversity measures of land use mixture compare the distributions

    of different land uses. We include the following measures: the Balance index,

    the Herfindahl-Hirschman index, the Dissimilarity index, the Gini coefficient,

    entropy, and the Atkinson index.

    Balance index

    Definition:The degree to which two different types of land uses (e.g., housing

    units and employees, or residential and non-residential land parcels) exist in

    balance to each other within a neighbourhood. If the two land use types are

    distributed evenly, the index is 1. The smaller the value, the greater the

    unevenness. If there is only one type of land uses in the neighbourhood, the

    index is 0.

    Unit of analysis:Neighbourhoods.

  • 8/11/2019 Mixed Land Uses White Paper

    18/39

    17

    The measures merit is its computational ease. However, MAUP is present

    because of the measure is based on aggregated units. For example, larger

    neighbourhoods will tend to have a higher jobs-to-residents balance.

    Herfindahl-Hirschman index (HHI)

    Definition: The Herfindahl-Hirschman Index (HHI index), a commonly

    accepted measure of market concentration used to detect market monopoly,

    can be used to assess the level of land use mixture. The HHI index is the sum

    of squares of the percentages of each type of land uses in the user-defined

    neighbourhoods. If there is only one land use type in the neighbourhood, HHI

    index will equal 10,000. The higher the value of HHI Index, the lower the

    level of land use mixture.

    Unit of analysis:Neighbourhoods.

    The main virtue of the HHI is its simplicity. However, it shares the same set of

    drawbacks with the measures of intensity as they all rely on the aggregated

    areal units for calculation.

    Dissimilarity Index

  • 8/11/2019 Mixed Land Uses White Paper

    19/39

  • 8/11/2019 Mixed Land Uses White Paper

    20/39

  • 8/11/2019 Mixed Land Uses White Paper

    21/39

    20

    Unit of analysis: Different levels of neighbourhoods (e.g., grids, census

    boundaries, or metropolitan areas).

    The Gini index is useful both to measure changes in distribution over time and

    for cross-sectional comparisons across neighbourhoods or metropolitan areas.

    As the D index, the Gini index is not a very discriminating indicator. Two

    very different distributions can have exactly the same Gini index. To report

    the Gini index for only one neighbourhood, by and large, is not sufficient to

    have a complete picture of the situation. It would be necessary to compare this

    value with the values obtained from the other neighbourhoods.

    Entropy measures

    Definition:The entropy index is a measure of variation, dispersion or diversity

    (Turner et al., 2001). It measures the degree to which land uses are

    heterogeneously distributed within a neighbourhood. A value of 0 indicates

    homogeneity, wherein all land uses are of one single type; a value of 1 means

    heterogeneity, wherein area is evenly distributed among all land use categories.

    Unit of analysis:Neighbourhoods.

    The entropy index incorporates more than two land use types in a single

    calculation, very conveniently aggregating a measure of land use diversity at

  • 8/11/2019 Mixed Land Uses White Paper

    22/39

    21

    various levels. Although other indices (e.g., dissimilarity) can be implemented

    to capture the integration of various land use types, the simplicity in

    computations of the entropy index makes it highly desirable.

    Atkinson index

    Definition: The Atkinson index (Atkinson, 1970), one of the few inequality

    measures that explicitly incorporates normative judgments about

    heterogeneous distribution, allows for the differential weights assigned to sub-

    units (e.g., grid cells within neighbourhoods) and thus enables grids where

    non-residential land uses are under- or over-represented to contribute more or

    less heavily to the overall index. The Atkinson index ranges between 0 and 1,

    with a score of 1 indicating the highest level of homogeneous land use

    distribution (or maximum segregation of land use types).

    Unit of analysis:Neighbourhoods.

    The Atkinson Index provides a practical opportunity for assigning weights to

    various land use distributions and making normative adjustments. For

    example, in a situation where some neighbourhoods have a large proportion of

    commercial land use areas due to the presence of strip malls, while some other

    neighbourhoods might only have small neighbourhood corner stores,

  • 8/11/2019 Mixed Land Uses White Paper

    23/39

    22

    researchers might value more the land use composition pattern in the latter

    neighbourhoods. Using the Atkinson index with an appropriate value for the

    inequality aversion parameter can accommodate these value judgements.

    3.3.2. Exposure

    Interaction Index

    Definitions: Exposure, originated in the field of measuring residential

    segregation, measures the degree of potential contact or possibility of

    interaction between two subject groups (Massey and Denton, 1988: 287). The

    interaction index measures the publicity of non-residential land uses to

    residential uses. Lower values of interaction indicate lower exposure.

    Unit of analysis:Neighbourhoods.

    Exposure and evenness (or diversity) measure different things: exposure

    measures depend on the relative sizes of the two groups being compared, while

    evenness measures do not (Massey and Denton, 1988). Exposure measures

    can thus correct for the problem (as we illustrated in Figure 2c) that evenness

    measures have.

    3.3.3. Clustering

  • 8/11/2019 Mixed Land Uses White Paper

    24/39

    23

    Clustering, originated in the studies of residential and income segregation,

    measures the extent to which areal units with different subjects adjoin one

    other, or cluster, in space (Massey and Denton, 1988: 293). We import one

    clustering measure to study the degree of spatial clustering of one type of land

    uses.

    Absolute Clustering

    Definition: Absolute Clustering summarizes the degree to which non-

    residential land uses are found in nearby as opposed to spatially distance areal

    units. The index ranges from 0 to 1, with higher values indicating a clustering

    of non-residential land uses.

    Unit of analysis:Neighbourhoods.

    Clustering considers the spatial arrangement of land uses within the

    neighbourhoods. Absolute Clustering corrects for the problem (as we

    illustrated in Figure 2a) that evenness measures have and can thus detect if the

    areal units with dominant one type of land uses are spatially clustered together.

    4. Empirical analysis of land use mixture measures

  • 8/11/2019 Mixed Land Uses White Paper

    25/39

    24

    In order to test the effectiveness of the measures on land use mix described in

    the previous section, we chose the City of Hillsboro which lies in the western

    portion of the Portland metropolitan area (See Figure 3) for an empirical study.

    We computed one measure of accessibility at the individual parcel level and

    ten measures of pattern at the neighbourhood level. We define neighbourhood

    by census blockgroup boundaries or by square gridcells -mile high and wide.

    We obtain the following GIS data:3 (1) Parcel-based (tax lot) property data;

    The parcel-based property data includes attributes for each parcel such as lot

    size, floor space, and information on land use type; (2) jurisdiction and census

    blockgroup boundaries; (3) Street networks, and (4) Parks, open space and

    other recreational land uses. As demonstrated in Figure 3, the larger scale of

    mixed activities including commercial strip, light-industrial (office), and multi-

    family residential land uses that were developed from after the World War II to

    the present day are agglomerated in the northeast corner of the city or along the

    arterial road, and most of the small-scale commercial enterprises, offices, and

    customary retail establishments that were developed prior to the war are in the

    downtown area of the city.

    --insert Figure 3 here--

    3These data are from Portland Metros Regional Land Information System (RLIS).

  • 8/11/2019 Mixed Land Uses White Paper

    26/39

    25

    The measure AG which is used to compute the accessibility of individual

    households to commercial stores replicates the actuality well (Figure 4). We

    see that housing units that are in the northeast corner of the city, closer to the

    downtown area, or along the major arterial roads, have higher accessibility.

    --insert Figure 4 here--

    For the measures of land use mixture between two groups (i.e., residential vs.

    non-residential land uses) at the blockgroup level, we compute the

    Dissimilarity index (DN), the Gini index, (GN), a set of Atkinson indices (A0.1,

    A0.5, and A0.9), the exposure indices (INT and ISO), and the Cluster index

    (CLUSTER). We provide a visual representation of the measures in Figure 5

    and the correlations among them in Table 1. The generalization of the indices

    suggests that the neighbourhoods in the northeast corner of the city, at

    downtown area, or along major arterial roads are more mixed than the other

    neighbourhoods.

  • 8/11/2019 Mixed Land Uses White Paper

    27/39

    26

    The evenness suite of measures (including DN , GN, A0.1, A0.5, and A0.9), not

    surprisingly, are highly correlated to each other (see Table 1). A0.5 is more

    similar to DN and GN since A0.5does not make adjustments to the under- or

    over-represented areal units. A0.1andA0.9modify the evenness by allowing the

    areal units where non-residential land uses are below- or above-average of the

    neighbourhoods proportion contribute more or less heavily to the overall

    indices. The exposure measure (INT) is correlated with the evenness measures.

    However, exposure measures are sensitive to the relative sizes of the two

    groups (i.e., residential vs. non-residential land uses) being compared and are

    thus able to detect that neighbourhoods aand b(see Figure 5), although score

    the same in the dimension of evenness since the distributions of land uses in

    the sub-units are comparable in relation to the larger blockgroups, do differ in

    the dimension of exposure. Since Neighbourhood bhas a larger proportion of

    non-residential land uses compared to neighbourhood a, the non-residential

    uses in neighbourhood bare less likely to interact with residential uses, as well

    as less likely to be isolated from other non-residential uses.

    TheClustering index taps into the spatial properties of adjacency or contiguity

    of non-residential land uses. For example, a smaller value of the CLUSTER

    index for Neighbourhood c than for Neighbourhood d (see Figure 5) reveals

  • 8/11/2019 Mixed Land Uses White Paper

    28/39

    27

    that the non-residential land uses in Neighbourhood care found in distant sub-

    units as opposed to nearby, while non-residential land uses in Neighbourhood

    dare more clustered.

    --insert Figure 5 and Table 1 here--

    To discover land use mixture between two groups (i.e., residential vs. non-

    residential land uses) within blockgroups, we experiment with two evenness

    measures at the -mile by -mile square level: DG and GG (see Figure 6).

    These two measures, within expectation, are performing alike and having a

    correlation of 0.81. A closer examination by comparing Figure 3 and 6

    suggests that, although the indices are effective in capturing intra-blockgroup

    variation in land use mixture, the outcomes are sensitive to the spatial position

    of the imposed grids.

    --insert Figure 6 here--

    For the measures of land use mixture among multiple groups at the blockgroup

    level, we compute the Dissimilarity index (D(m)), the Entropy index, (E2), the

  • 8/11/2019 Mixed Land Uses White Paper

    29/39

    28

    Herfindahl-Hirschman Index (HHI), and a set of Atkinson indices (A(m)0.25,

    A(m)0.5, andA(m)0.75). We present the results of this set of indices in the lower

    panels of Figure 6 and the correlations among the indices in Table 2. An

    overview of the indices suggests that they correspond to the findings of the

    two-group measures: the neighbourhoods in the northeast corner of the city, at

    downtown area, or along major arterial roads are more mixed than the other

    neighbourhoods. The values of the Entropy, HHI, and Atkinson family of

    indices which highly correlate with each other point to the same generalization.

    --insert Table 2 here--

    5. Discussion and Conclusions

    Measures of land use mix are useful for understanding the patterns of land use

    distribution. They also enable researchers to evaluate their relationship with

    individual and community outcomes in disciplines such as including landscape

    ecology and the environment (air quality, water quality), transportation (auto

    ownership, travel behaviour), health outcomes (physical activity behaviour,

    obesity), and housing markets (property values). Despite the burgeoning

    interests in studying mixed land uses and their consequence there is a paucity

  • 8/11/2019 Mixed Land Uses White Paper

    30/39

    29

    of research on the measurement of such mixed land use. In this paper we

    provided a synthetic examination of an array of land use mix measures which

    would tap various dimensions of the urban land use mixture. We classified

    existing indices as measures of accessibility, intensity and pattern. With the

    purpose of evaluating the measures, we also applied selected measures in an

    empirical case study.

    Measures of accessibility are valuable for directly incorporating geographic

    distance into the measure. The distance measures involve unsophisticated

    computation and provide convenient information on individual units or

    neighbourhoods accessibility to mixed land activities. They range in

    sophistication and computational burden from simple measures (e.g., distance)

    to measures requiring parcel-level data and calibration of the parameters (e.g.,

    gravity with competition and destination choice measures). Their conceptual

    simplicity, coupled with the requisite disaggregate-level data make these

    measures comprehensive and suitable for studies focusing on individual (as

    opposed to community) outcomes.

  • 8/11/2019 Mixed Land Uses White Paper

    31/39

    30

    Measures of intensity can only be implemented at aggregated unit level and

    entail the least amount of computation and data requirements. Because these

    measures ignore information on the spatial configuration of land uses, they can

    be considered aspatial. Their major strength, relative to all other measures, is

    the conceptual and computational simplicity. This translates into ease of

    communicability. Their strength, however, also is their major weakness. Our

    review highlighted concerns related to the reliance on an aggregate analysis

    unit (such as the modifiable aerial unit problem, edge effects, and issues with

    the scale of analysis).

    Measures of pattern are more adequate for capturing the diversity, isolation,

    and the clustering of land uses. Our correlational analysis demonstrates a high

    degree of interrelatedness among our diversity measures within pattern.

    Among the measures of evenness, the two-land use type Dissimilarity index is

    valuable for its ease of interpretation and computation. It correlates highly

    with other measures (e.g., the Gini and the Atkinson indices) but requires less

    computational and data management burden. The Dissimilarity indexs

    usefulness, however, is limited to the evaluation of evenness between two

    groups of land uses. The data management complexity of our multi-group

    implementation of the dissimilarity index limits its usefulness for practitioners.

  • 8/11/2019 Mixed Land Uses White Paper

    32/39

    31

    Because the Entropy and HHI indices can handle multiple land use types using

    relatively simple calculations, we consider them as convenient measures of

    land use diversity. The empirical comparison of these indices suggests that the

    HHI may be easier to communicate to a non-technical audience, although the

    Entropy index has deeper roots in the literature and can have a behavioural

    interpretation. Finally, the Interaction index and the Clustering index are

    complements to measures of evenness and diversity. They contribute

    information about clustering and thus are valuable for providing a richer

    depiction of the land use distribution in a given area. The effectiveness of

    these two measures is, however, constrained to two land use types.

    Our review of the land use mix measures, and the empirical application, offer

    an improved understanding of the measures properties for researchers and

    practitioners. It is tempting to ask which measure is the most appropriate one

    in evaluating land use mixture. Obviously, there is no single best measure of

    land use mixture, since each measure captures different dimensions of how

    land uses are distributed in space. However, our review and the empirical

  • 8/11/2019 Mixed Land Uses White Paper

    33/39

    32

    application can provide insights for researchers and practitioners regarding the

    appropriateness of particular measures for particular purposes.4

    First, the choice of measure is depended on the extent to which a measure

    captures the presence and configuration of land uses in space. For example, is

    thepatternof several land uses more or less of interest than the mere presence

    of those uses in the study area? Should the measure account for more than two

    land use types? Will the index measure what the researcher or practitioner

    wants to measure?

    Second, practical considerations should also influence the choice of measure.

    These include data collection and management, computational burden, and

    ease of communicability. While some measures require data manipulations that

    require database programming, others result naturally from a land use cover

    map. The technical appendix containing the various implementations of

    measures confirms the importance of practical considerations in deciding

    which measures to use. By most accounts, relatively simple measures have

    been implemented more frequently than complex measures. Of course, this

    4Others have relied on the mathematical properties of selected measures discussed here, but in

    the context of racial segregation (James and Taeuber, 1985).

  • 8/11/2019 Mixed Land Uses White Paper

    34/39

    33

    simplicity has tradeoffs and may contribute to explain divergent results in

    various disciplines with respect to the relevance of land use mixtures for

    community and individual outcomes.

    Finally, and perhaps most importantly, the connection between the measures

    and the purpose of the investigation should drive the measures selected. In

    other words, measures should be selected based on the substantive questions

    driving the inquiry. If the question being asked is about non-motorized travel

    behaviour then the location of commercial and office land uses relative to

    residential uses is of paramount interest. A two-land use type measure may

    suffice. By contrast, if the question motivating the research is the impact of

    non-residential land uses on property values, then the location of at least parks,

    industrial and commercial uses relative to residential units should be of

    concern.

    There is undoubtedly a need to acknowledge that the land use mixture is only

    one, perhaps modest, influence on the individual, neighbourhood, and societal

    outcomes. Nevertheless, the measures of mixed land uses are useful for

    quantifying the distribution of land uses that can be used across disciplines to

  • 8/11/2019 Mixed Land Uses White Paper

    35/39

    34

    investigate, through empirical research, the new inquiry on the importance of

    the impact of the land use mixture on a variety of outcomes. Our exercise is a

    contribution to the investigation, and an attempt to begin a long-term process

    of refinement and advancement in this field.

  • 8/11/2019 Mixed Land Uses White Paper

    36/39

    35

    References

    Atkinson, A B, 1970, "On the Measurement of Income Inequality"Journal ofEconomic Theory2: 244-263

    Brown, M, 1994, "Using Gini-style indices to evaluate the spatial patterns of

    health practitioners; theoretical considerations and an application based

    on the Alberta data" Social Science and Medicine389:1243-1256

    Centres for Disease Control and Prevention (2005) Active CommunityEnvironments http://www.cdc.gov/nccdphp/dnpa/aces.htm, 2005 [cited

    February 28 2005]

    Cervero, R, 1996, "Mixed land-uses and commuting: evidence from the

    American Housing Survey" Transportation Research A305:361-377

    Cervero, R and Duncan, M, 2003, "Walking, bicycling, and urban landscapes:Evidence from the San Francisco Bay area" American Journal of

    Public Health939:1478-1483

    Cervero, R and Duncan, M, 2004, "Neighbourhood composition and

    residential land prices: Does exclusion raise or lower values?" Urban

    Studies412:299-315

    Cervero, R and Kockelman, K, 1997, "Travel demand and the 3Ds: Density,

    diversity and design" Transportation Research D23:199-219

    Duany, A and Plater-Zyberk, E, 1992, May, "The second coming of theAmerican small town" Plan Canada: 613.

    Duncan, O D and Duncan, B, 1955, "A methodological analysis of segregationindexes"American Sociological Review20: 210-217

    Ewing, R, Pendall, R and Chen, D, 2003, "Measuring sprawl and its

    transportation impacts" Transportation Research Record1831:175-183

    Frank, L D, Andresen, M A and Schmid, T L, 2004, "Obesity relationships

    with community design, physical activity, and time spent in cars"American Journal of Preventive Medicine272:87-96

  • 8/11/2019 Mixed Land Uses White Paper

    37/39

    36

    Frank, L D, Schmid, T L, Sallis, J F, Chapman, J and Saelens, B E, 2005,

    "Linking objectively measured physical activity with objectively

    measured urban form" American Journal of Preventive Medicine 282S2:117-125

    Frank, L D, Stone, J, Brian and Bachman, W, 2000, "Linking land use withhousehold vehicle emissions in the central puget sound: methodological

    framework and findings" Transportation Research Part D: Transport

    and Environment53:173-196

    Geoghegan, J, 2002, "The value of open spaces in residential land use" Land

    Use Policy191:91-98

    Geoghegan, J, Wainger, L A and Bockstael, N E, 1997, "Spatial landscapeindices in a hedonic framework: an ecological economics analysis

    using GIS"Ecological Economics233:251-264

    Giles-Corti, B, Broomhall, M H, Knuiman, M, Collins, C, Douglas, K, Ng, K,

    Lange, A and Donovan, R J, 2005, "Increasing walking: How

    important is distance to, attractiveness, and size of public open space?"American Journal of Preventive Medicine282, Supplement 2:169-176

    Giles-Corti, B and Donovan, R J, 2002a, "The relative influence of individual,

    social and physical environment determinants of physical activity"

    Social Science & Medicine5412:1793-1812

    Giles-Corti, B and Donovan, R J, 2002b, "Socioeconomic Status Differences in

    Recreational Physical Activity Levels and Real and Perceived Accessto a Supportive Physical Environment" Preventive Medicine356:601-

    611

    Gini, C, 1912, "Variabilit e mutabilita" reprinted inMemorie di metodologica

    statistica (Ed. Pizetti E, Salvemini, T). Rome: Libreria Eredi VirgilioVeschi 1955.

    Greenwald, M J and Boarnet, M G, 2001, "Built environment as determinant of

    walking behaviour: Analyzing nonwork pedestrian travel in Portland,

    Oregon" Transportation Research Record1780:33-42

    Handy, S, 1996, "Urban form and pedestrian choices: study of Austinneighbourhoods" Transportation Research Record1552:135-144

  • 8/11/2019 Mixed Land Uses White Paper

    38/39

  • 8/11/2019 Mixed Land Uses White Paper

    39/39

    Sallis, J, Johnson, M, Calfas, K, Caparosa, S and Nichols, J, 1997, "Assessing

    perceived physical environmental variables that may influence physical

    activity"Research Quarterly for Exercise and Sport684:345-351

    Shen, Q, 1998, "Location characteristics of inner-city neighborhoods and

    employment accessibility of low-income workers" Environment andPlanning B25: 345 365

    Song, Y and Knaap, G J, 2004, "Measuring the effects of mixed land uses on

    housing values" Regional Science and Urban Economics 34 6:663-680

    Song, Y and Sohn, J, In press, " Valuing spatial accessibility to retailing: A

    case study of the single family housing market in Hillsboro, Oregon"

    Journal of Retailing and Consumer Service

    Srinivasan, S, 2002, "Quantifying spatial characteristics of cities" UrbanStudies 39 11:2005 -- 2028

    Stokols, D, 1992, "Establishing and maintaining healthy environments:Toward a social ecology of health promotion" American Psychologist

    471:6-22

    Turner, M G, 1990, "Spatial and temporal analysis of landscape patterns"

    Landscape Ecology41:21-30

    Turner, M G, Gardner, R H, and ONeill, R V, 2001. Landscape Ecology in

    Theory and Practice: Pattern and Process.Springer Verlag, New York.

    Weber, J and Kwan, M, 2003, "Evaluating the effects of geographic contexts

    on individual accessibility: A multilevel approach" Urban Geography

    248: 647671

    Weibull, J W, 1976, "An axiomatic approach to the measurement ofaccessibility"Regional Science and Urban Economics6: 357 379

    Wong, D W S, 2003, "Implementing spatial segregation measures in GIS"

    Computers, Environment and Urban Systems27: 53-70