Measurement and Analysis of Livability
The second chapter emphasized that livability is a spatial and temporal phenomenon. This chapter discusses some of the issues involved in measuring and analyzing livability, including how to measure place-based indicators. Place-based indicators (and indeed any place-based measurements) involve issues such as the effects of arbitrary geographic boundaries and units, the possibility of ecological fallacy, deciding when measurement should occur, reconciling incompatible data units, and considering spatial data in statistical methods.
Issues involved in measuring accessibility to opportunities and to resources are also discussed. Individual accessibility to opportunities and resources is a central component of livability. However, “accessibility” is a multifaceted concept involving some challenging measurement issues, for example, space-time accessibility measures. These measures derive from the time geographic perspective discussed in Chapter 2 and capture the effects of individual activity schedules on accessibility. Since daily and weekly activity schedules vary widely by socioeconomic variables such as class, life cycle, culture, and gender roles, space-time accessibility measures are sensitive to individual differences in accessibility. Space-time accessibility measures can support livability measures that take into account the varying access to resources and opportunities between social and demographic groups in a community. A case study in Box 3.1 describes the planning of a national monument area in southern Utah, which allowed diverse groups to access geospatial data that provided information needed to fully participate in the monument planning process.
mental groups were most proficient at taking advantage of this information. They thought that access to data for analysis much improved their ability to make effective comments on the draft plan.
The general public mostly responded to maps showing roads and plan alternatives. Citizens were able to use their own knowledge of the area to comment on the appropriateness of specific plans. In many cases, their comments filled in gaps in the knowledge of the planning group; not all useful data are available to governments trying to achieve a successful planning effort. In other cases, seeing plan details diffused fears that the general public had about the loss of access to favorite sites within the monument.
As a result of the public input, the planning team added and removed several roads from the preferred transportation alternative coverage based on map-driven comments from the public comment period. Administrative roads in the accepted alternative were reduced from 310 miles in the draft plan to 192 miles in the proposed plan. In addition, changes and buffer zones were added to the monument management zones and boundaries based on public comment. More than 6,800 comments were received regarding the draft plan.
A qualitative analysis of the process found the following benefits: increased participation in the planning process, increased understanding of the plan, more substantive comments, improved communication, improved geospatial database for the monument, and an improved proposed plan. The paper and electronic GIS maps increased citizens’ understanding of the plan. Individuals were able to get a clear picture of the process that led up to policy decisions. In addition, GIS use increased among stakeholder groups as a result of this data sharing pilot project.
GIS maps improved the planning process by providing stakeholders with a common language—GIS allowed them to discuss issues, rather than dispute location of features. The public found visual information easier to understand than written chapters. Individuals found that the increased perspective on the implications of various alternatives clarified their initial ideas about the plan. See examples of these collaborative GIS efforts on Plates 3, 4, and 5
SOURCE: BLM (1999).
DEVELOPING PLACE-BASED INDICATORS
Most place-based analyses use data reported at an aggregate level for some kind of geographic area. Examples include census tracts, census block groups, traffic assignment zones, school districts, or political units such as municipalities and counties. It is not always the case that these “administrative” areas match well with the definition of places as described in the previous chapter. These areas also may not adequately
represent characteristics or needs of individuals. A discussion of some of the problems associated with measuring and analyzing the attributes of places follows.
Arbitrary Geographic Boundaries
Geographic boundaries created for measurement or administrative purposes can create misleading spatial patterns in geographic phenomena. Trying to place an external boundary around a study region can create two artificial effects in measurement and analysis. One is an “edge effect” created by ignoring interdependences that occur outside the bounded region. A second effect is the artificial shape imposed by the boundary. Shape can affect the measurement of spatial point patterns (e.g., reported crime locations) since these compare the points’ locations relative to area. For example, as spatial units become more elongated, point pattern statistics tend to report higher levels of clustering for the same point pattern within that unit (Fotheringham and Rogerson, 1993).
Shape relative to area can also affect the measurement of interactions (e.g., origin-destination flows) since these are often recorded only when they cross an artificial boundary. Information about shape and area can be exploited to more accurately estimate distances from travel surveys (Rogerson, 1990) or to locate traffic counters, travel survey stations, or traffic monitoring systems (Kirby, 1997).
The problem of defining “urbanized areas” provides a relevant example of geographic bounding problems. The U.S. Census defines urbanized areas as jurisdictions with 1,000 persons or more per square mile. Figure 3.1 illustrates U.S. Census urbanized area boundaries for a portion of the Rocky Mountain Front Range that includes Ft. Collins and Greely, Colorado. Purple lines indicate the urbanized area boundaries, red shading indicates urban land use, orange shading indicates suburban land use; and yellow shading indicates exurban land uses. As can be seen, the census definition of urbanized area is problematic. Similarly, urban livability indicators such as measures of sprawl often ignore interdependences and interactions with proximal (or nearby) rural areas (Theobald, 2001).
There are several strategies for resolving geographic boundary problems in measurement and analysis (see Griffith, 1983; Griffith and Amrhein, 1983; Martin, 1987; Wong and Fotheringham, 1990). A practical computational strategy is to use GIS tools to manipulate boundaries systematically and to conduct the measurement and analysis given these different boundaries. This provides a sensitivity analysis of the indicator with respect to boundary definitions. Without this type of sensitivity analysis, the reliability and robustness of place-based livability measures that rely on administrative boundaries are unclear.
Arbitrary Geographic Units
Closely related to artificial boundaries is the problem of the effect of arbitrary geographic units on place-based measurement and analysis. Data for livability indicators are often spatially aggregated according to a defined spatial zoning system such as census tracts, census block groups, school districts, or political units such as municipalities or counties. These units can be meaningful in reality; for example, municipalities correspond
to geographies of taxation and service provision. Environmental regions, such as watersheds, can be identified easily and bounded, allowing some physical variables to be measured nonarbitrarily. However, the “coarseness,” created by spatially aggregated reporting units is often a barrier to understanding the spatial variation of many important social variables. The problem arises when measuring both the average level of a variable and its unequal distribution over the population.
Problems associated with arbitrary geographic units are known as the modifiable areal unit problem (MAUP) in the spatial analysis and quantitative geography literatures. The MAUP occurs when analyzing data that are recorded (or reported) for arbitrary spatial units. If the spatial zoning system is arbitrary or “modifiable,” then the results of any measurement or analysis based on those units are also arbitrary or modifiable (see Miller, 1999a).
The MAUP has two dimensions. One dimension is scale: this relates to the level of spatial aggregation in the data. For example, in a multicounty metropolitan area, we might have data for each city, town, and township, or we might have only an average for each county. The other dimension is zoning, which refers to changes in the spatial partitioning given a fixed level of spatial aggregation (Openshaw and Taylor, 1979; Wong and Amrhein, 1996). For example, we might have data that show the averages for groupings such as center city and inner and outer suburbs, or data showing averages for central city and eastern and western suburbs, where these groupings are all roughly equal in size.
The MAUP can be illustrated using an example based on Monmonier (1996, pp. 140-145). Figure 3.2 is a map of a region of 16 towns that vary in population size. Assume that Towns 4, 10, and 13 are considerably larger in population than the rest. Also assume that a livability index has
been calculated for each town based on an average value per household. This index would be based on factors agreed upon by the communities involved as indicators of livability. For the purposes of this discussion, it does not matter which indicators or set of indicators was chosen by these towns for this comparison. Figure 3.3 maps the livability index for each town. There are three levels of the index: (i) low—index varies between 8 and 12; (ii) medium—index varies between 18 and 22; (iii) high—index varies between 28 and 33.
Note that the sharp differences between low, medium, and high categories means that a large change in a town’s index is required to change its category. This example shows that the MAUP is independent of the imprecision in choosing the dividing lines between low, medium, and high. The spatial pattern of livability in Figure 3.3 shows a general north-south variation. This approach is not perfect, however; Towns 1 and 15 are notable departures.
Figure 3.4 shows the calculated livability categories when grouping towns into three north-south regions. The spatial pattern of livability now shows an exact north-south variation, hiding the fact that several towns are not typical of their regions. For example, livability seems to jump two levels in Town 1 and to drop two levels in Town 15. Figure 3.5 maps the index for a different grouping into three east-west regions. This grouping creates even more dramatic change in apparent livability: 11 of the 13 towns have different classifications relative to those in Figure 3.3. Town 16
to each other, they show the effect of zoning since both have the same degree of aggregation (two four-town groupings and one eight-town grouping) yet give different pictures of livability in many towns.
There are many other possibilities for zoning and aggregation, each of which can create different representations of livability. For example, if all 16 towns are made into a single group (perhaps to compare them to another region not shown here), the single group would have a medium level of livability based on the index. As another example, Figure 3.6 shows the results of grouping only three towns into a single metropolitan area of one large town (Town 10) plus two smaller ones (Towns 14 and 15). In this case, the medium rating of the special area overstates the livability index for Town 14 and understates it for Town 15 but is accurate for Town 10. While this hypothetical example and the maps illustrate the point for towns, livability can be affected strongly by smaller-scale changes. The livability of two neighboring houses on the same street could be assessed very differently, or whole areas of a city might be assessed as having a low livability index without any discussion of how to change these factors. Appropriate treatment of scale is also important for smaller areas such as neighborhoods and, indeed, for areal units of any size.
The simple examples in Figures 3.2 through 3.6 illustrate that place-based livability measurements are partially an artifact of the spatial reporting units. We can create substantially different patterns of livability
from the same data, simply by changing the spatial units. This strategy brings into serious question the reliability and robustness of any place-based livability measurement that uses spatially aggregated data.
What can be done about the modifiable areal unit problem when measuring and analyzing livability? One strategy is to attempt to develop optimal spatial units for the particular problem being addressed. This requires assembling smaller spatial units into larger spatial units in order to optimize some criterion (e.g., equal proportion of minority groups). Horn (1995), Openshaw (1977, 1978), and Openshaw and Rao (1995) discuss methods for optimal spatial unit design. The multidimensional nature of livability creates some problems, since optimal spatial unit design becomes more difficult, and some cases became impossible as the number of variables increases (although correlations among variables can reduce this complexity).
A second strategy is to assess the impacts of the arbitrary zoning system on the measured livability indicators. This is similar to the sensitivity analysis strategy discussed with respect to arbitrary geographic boundaries. GISs can support sensitivity analysis of MAUP effects in livability analysis through their ability to reorganize spatial data as well as visualize and communicate results (Fotheringham and Rogerson, 1993). Fotheringham and Wong (1991) provide a good illustration of sensitivity analysis with respect to spatial aggregation and zoning.
Another issue that is related to problems with aggregate geographic units is the ecological fallacy. The ecological fallacy occurs when the characteristics of individuals are inferred from aggregate data. For example, it can be tempting to conclude that a livability attribute calculated from aggregate data for a region, city, census tract, or census block reasonably describes the situation of most individuals within that geographic unit. In fact, of course, aggregate measures can mask a great deal of variability among individuals (Knox, 1978). This danger is common to all analyses with aggregate data, whether spatial or nonspatial.
It is difficult to argue that either a place-based or a person-based perspective is more appropriate. Each perspective provides a different lens through which to view livability. For example, using a person-based perspective alone may lead us to commit an individualistic fallacy by ignoring holistic factors that emerge at aggregate levels (Johnson, 1986). Both place-based and person-based perspectives are necessary for appropriate analysis of livability. For example, a pure individualistic perspective ignores the social networks and informal social capital that can develop in some neighborhoods.
Time of Measurement
The time geographic perspective discussed in the previous chapter tells us that human settlement landscapes are affected by the timing of events as much as by the spacing of people, activities, and structures.
We can analyze space-time human geographies from both people-based and place-based perspectives. In a people-based strategy, we analyze and visualize individual space-time paths, trying to understand individual movements and constraints on movements. The activity diary data collected for many years by local metropolitan planning organizations and departments of transportation can support livability analysis from that perspective. These data track individual movements and activities in time and space in order to plan transportation infrastructure and formulate transportation policy that are sensitive to the required and desired activities that comprise individuals’ lives. The development and deployment of position-aware technologies such as in-vehicle navigation systems, cellular telephones, personal digital assistants, and wireless Internet clients are lowering the cost required to conduct detailed studies of the space-time human geographies of cities (Smyth, 2001).
GISs can be used to visualize and explore these space-time trajectory data. We can visualize time-space paths in a static mode using a three-dimensional model of space-time that shows the entire path within a geographic space and a fixed domain of time (Forer, 1998). Another possibility is animated maps that show the dynamic evolution of these paths in geographic space over time (van der Knaap, 1997; Kwan, 2000a). The increasing possibilities for collecting space-time path data in real time through intelligent transportation systems, location-based services, and other position-aware technologies mean that animation and other exploratory geographic visualization techniques will become essential for understanding space-time path data.
A place-based strategy is to aggregate the space-time paths into space-time units using temporal measurements taken at discrete time intervals for the spatial reporting units (Taylor and Parkes, 1975). Examples of space-time units are census block groups with attributes measured at different representative times such as “weekday morning,” “weekday evening,” “weekend morning,” and “weekend evening.” In contrast, the standard census practice of recording attributes based on home location assumes a particular time (e.g., weekday evening) that may not accurately represent the unit over the daily and weekly clocks. A central business district can appear relatively empty based on this measure. Also, neighborhoods with significant social, cultural, and entertainment activities, such as restaurants and nightclubs, may have completely different demographics on weekend evenings than during most other time of the week.
We can analyze the aggregate data using statistical techniques (while being mindful of arbitrary boundary and unit problems, as well as dependences in the data). The space-time units can show diurnal patterns in the social geography of the city as well as interactions between activities, social settings, and urban form. They allow measurement and prediction of the impact of changes in demographics, socioeconomic structures, and activity patterns within the urban environment as well as time-varying demands for transportation infrastructure. Longitudinal studies can allow assessment of the effect of long-term changes on livability; for example, the effects of changing demographics, continuing intensive use of the automobile, the growth of multi-income households and participation of women in the labor force, and the wider use of telecommunication technologies (Goodchild and Janelle, 1984; Janelle et al., 1998).
Incompatible Data Units
Livability and sustainability are complex phenomena measured across multiple dimensions. Geographical data are often collected using different, sometimes arbitrary, spatial units. For example, data available from a census, based on census tracts or other secondary sources, may not match the traffic analysis zones used by transportation planners in a transportation department or a metropolitan planning organization (MPO). Integrating these data means resolving different spatial recording systems. Spatial basis transfer is the term used to describe the conversion of data from one spatial system to another. If spatial basis transfer is not conducted properly, the resulting place-based measurements can be arbitrary, and even misleading, since the geographic framework for the measure is distorted.
Sometimes there is a need to transfer data from one area (the source zone) and apply them to another (the target zone). If the target zones nest perfectly within the source zones, this process is straightforward. If the nesting is not perfect, then areal interpolation is required (Goodchild and Lam, 1980). The appropriate method for interpolating data from source zones to target units depends on beliefs or assumptions about the spatial variation of the data within these zones (Goodchild et al., 1993). If both the source and the target zones are relatively homogeneous, the method of areal weighting can be used. This method distributes the data in the source zones to the target zones based on the share of the source zone within each target zone (see Goodchild and Lam, 1980). If the source zones are not homogeneous and there are other data that say something about the distribution of the variable within each zone (e.g., housing value as a surrogate for household income), statistical techniques such as the expectation-maximization algorithm can be used (see Flowerdew et al.,
1991). If neither the source zones nor target zones are homogeneous but we have access to a third set of zones with a surrogate variable that has a constant density, these control zones can be used in an intermediate stage of the areal weighting technique to interpolate in two steps, first to the control zones and then to the target zones (Goodchild et al., 1993).
Imagery derived from remote sensing platforms is an increasingly viable source of socioeconomic as well as physical data. Traditional satellite-based remote sensor systems were limited to spatial resolution no higher than 10 meters. New high-resolution sensing systems can achieve spatial resolutions 1 meter higher. Spectral resolution is also improving: new hyperspectral sensor systems such as the Airborne Visible InfraRed Imaging Spectrometer (AVIRIS) capture more than 200 very narrow bands, providing a detailed spectral signature that allows discrimination to the subpixel level (i.e., the groundcover “mix” within a pixel). Jensen and Cowen (1999) discuss minimum spatial, temporal, and spectral resolutions required in remote sensing systems to extract urban and suburban infrastructure. Mesev, et al. (1996) discuss methods for inferring urban socioeconomic data from remote sensing imagery.
Spatial-Temporal Data and Inferential Statistics
The connectedness of livability in space and time is concerned with two other issues related to inferential statistics: namely, spatial dependence and spatial heterogeneity. We often refer to spatial dependence as spatial autocorrelation when discussing this property from a statistical perspective. Spatial dependence refers to the tendency of individuals or geographical units that are proximal in space to exhibit similar characteristics. Closely related is temporal dependence or temporal autocorrelation.
Spatial heterogeneity relates to the inadequacy of overall (system-wide) parameters in describing a specific phenomenon at individual locations. Spatial heterogeneity can occur for two reasons (Fotheringham, 2000). One reason is that some relationships are intrinsically different across space; for example, people’s behavior may vary by community or administrative, political, economic, and other boundaries or contexts. This creates contextually different responses to the same stimuli. Measuring spatial heterogeneity is a precursor to more intensive study to identify these contextual effects. Another reason is that the statistical model is not specified properly; one or more variables are missing or do not have the correct functional form. This statistical model can lead to misleading conclusions from the model. A classic example is a disaggregate spatial interaction (“gravity”) model leading to the conclusion that people in Albany, New York, are “jet-setters” compared to those in Los Angeles, California (Fotheringham, 1981). In this case, we must capture the spatial heteroge
neity in the model to account for the missing or incorrectly specified effects.
Multivariate statistical techniques such as regression analysis are often used to test causal relationships between livability indicators and fiscal, social, economic, and environmental variables. These methods can be used to determine how much of the variability is attributable to specific factors. Standard multivariate statistical methods make the assumption that all observations are independent of one another, that is, they do not vary one with another. With geographic data, independence cannot always be assumed because of spatial dependence, whereby factors do vary in relation to one another. Spatial dependence in the observations means that parameter estimates and significance tests are unreliable (Anselin and Griffith, 1998). It does not necessarily affect the model’s predictive accuracy but does seriously undermine the ability to use calibrated parameters to explain the relative causal effects of the independent variables.
There are many different methods for dealing with the challenges of measuring spatial dependence and spatial heterogeneity (see Getis and Ord, 1992; Anselin, 1995; Ord and Getis, 1995). Problems associated with spatial dependence among observations in multivariate regression and related techniques can be resolved by including spatial autocorrelation in the dependent variable, independent variables, error terms, or some combination (Anselin, 1988, 1993). Spatial dependence and spatial heterogeneity can be captured simultaneously using geographically weighted regression. Geographically weighted regression generates disaggregate, location-based regression parameters that show spatial variations in the relationships between the independent variables and the dependent variable (see Brunsdon et al., 1996). Geographically weighted regression results are easily mapped, creating powerful geographic visualizations to highlight spatial trends and spatial variations, and to identify local exceptions to these relationships (Fotheringham, 2000).
Accessibility is a key component of livability that implicitly or explicitly underlies many measures and analyses of livability. Accessibility is also closely intertwined with policies that intentionally or unintentionally influence livability.
Many livability measures assume that the resources and opportunities at a place are perfectly available to individuals who are “proximal” to that location. New policies that attempt to influence livability also make this assumption. However, factors other than propinquity can affect the ability of individuals to obtain resources and opportunities. This result means that measures can overestimate livability and the effectiveness of
related policies by masking individual variations in the benefits actually obtained from resources and opportunities.
Since accessibility is central to urban theory and policy, there is a long history of attempts to measure this concept. Accessibility can be based on potential or on outcomes (Scott, 2000). Potential measures attempt to quantify the ability of locations or individuals to interaction with other locations or individuals. Examples include time model-based (“isochrones”) and spatial interaction (“gravity”) model-based measures, such as the well-known Hansen potential measure and its derivatives (Hansen, 1959; Geertman and van Eck, 1995):
where ai is the accessibility of location i; oj is the attractiveness of opportunities at destination j; dij is the distance (or travel time) between locations i and j; and ß is a calibrated “friction-of-distance” parameter. Out-come measures use actual travel behavior and interactions to quantify “realized accessibility” as a surrogate for accessibility.
Another issue is that of distinguishing between accessibility and mobility (Scott, 2000). Mobility-based measures simply quantify mobility or the physical ease of movement within a given environment. These measures include travel times or distance. Broader conceptualizations of accessibility treat mobility as only one component of a wider context for travel that includes the opportunities at travel destinations and the general costs (social, economic, political, psychological) of reaching those destinations (Handy, 1994).
Accessibility measures involve implicit assumptions regarding what is being accessed, by whom, and how. Accessibility measures should be sensitive to the widely varying needs and resources of different social and demographic groups. The daily, weekly, and monthly activity schedules of individuals vary substantially by socioeconomic class, life cycle, culture, and gender roles (see Golledge and Stimson, 1997). Accessibility measures that are sensitive to different social and demographic contexts should incorporate the spatial and temporal constraints imposed by individuals’ activity schedules and the ability to overcome these spatiotemporal constraint results.
Space-time accessibility measures are measures that incorporate constraints imposed by individuals’ activities in space and time. Space-time accessibility measures can capture these constraints effectively (Kwan,
1998; Miller and Wu, 2000). A central concept in space-time accessibility measures is the space-time prism (Figure 3.7). This figure is an extension of the space-time path discussed in Chapter 2 (Figure 2.1). In this simple example, an individual is required to be at a specific location until a specified time and then return to that location at a later time (for example, a person who can leave the office during lunch but must return for the afternoon). Given this anchoring location, a time “budget” for travel and activity participation, and an assumed average travel velocity that is uniform across space, a three-dimensional space-time prism can be constructed. The interior of the prism is the potential path space (i.e., all locations in space and time that can be occupied by the space-time path during that discretionary period). The projection of the prism to the two-dimensional plane provides the potential path area (i.e., all locations in geographic space that the person can occupy during that discretionary period). Figure 3.7 is a simple illustration; the space-time prism can be
more complex geometrically with noncoincident anchoring locations, different spatial metrics, and required activity time removed from the prism (see Burns, 1979).
The classical space-time prism assumption of a uniform travel velocity is a glaring oversimplification of more complex travel environments where travel velocities can vary by location (e.g., central city versus suburb, residential street versus highway) and time (e.g., peak hours versus non-peak hours). The greater availability of digital geographic data and the increased ability to process geographic information can allow one to relax this assumption when calculating and applying space-time prisms. One possibility is to use travel time data for transportation networks to construct network versions of the space-time prism. A simple algorithm based on the shortest-path procedure allows calculation of a network-based potential path tree: this shows all nodes in the network that are reachable given anchoring locations, a time budget, and travel times within the network (Miller, 1991). Behavioral constraints such as limited information can also be included (Kwan and Hong, 1998). Quantitative accessibility measures, such as constrained potential measures, can also be calculated using these and other network space-time prism measurements (Miller, 1999b; Miller and Wu, 2000).
Accessibility is an important component of livability, and more specifically of social equity as it relates to livability. However, in terms of analysis, accessibility is a complex function of distance, time, ease of mobility, and other factors. Space-time accessibility measures derived from time geography highlight the role of transportation technology in trading time for space but do not incorporate the role that communication and information technologies play in eliminating space for certain activities. Yet, even as these technologies permit more activities and information exchange in cyberspace, persistent inequalities in access to information technologies (often called the digital gap or divide) will create even wider differences in accessibility among social and demographic groups (Dodge and Kitchin, 2001). Researchers are extending the analyses of time geography to include communication and information technologies using time as a common metric to integrate geographic and cyberspace (see Adams, 1995, 2000; Kwan, 2000b; Shen, 2000). These analyses should create powerful, integrated measures of accessibility that capture the use of (or exclusion from) transportation and information technologies within the constraints dictated by activity schedules and locations.
Space-time perspective offers a powerful perspective for measuring accessibility at an individual level. Implementing this perspective in applied analysis requires data on individual space-time activities. In the past, collecting these data was prohibitively expensive, time-consuming, and fraught with errors. However, the increasing deployment of position-
aware technologies, such as cell phones, wireless personal digital assistants, and global positioning system-enabled devices, is greatly lowering the cost and improving the accuracy of these data (see Smyth, 2001). Theories and tools for analyzing these data are also becoming increasingly available; for examples, consult the reference cited previously in this section as well as the edited volume by Frank, et al. (2001).
SUMMARY AND CONCLUSION
This chapter discusses spatial and temporal issues involved in measuring and analyzing livability. The discussion includes how to measure place-based indicators and how to measure accessibility, a complex phenomenon that conditions livability. Major conclusions and recommendations follow:
Many geographic boundaries are arbitrary and affect the collection of geographic data and the measurement of livability. Digital geographic data and GIS tools should be used to conduct sensitivity analyses of livability indicators with respect to boundary changes.
Many aggregate geographic units are arbitrary and create artificial effects in data collection and livability measurement with respect to spatial aggregation and zoning. Digital geographic data and GIS tools should be used to conduct sensitivity analyses of livability indicators with respect to changes in aggregation and zoning. Computational methods can also be used to form optimal spatial units for some measures.
Using only a place-based perspective may result in ecological fallacy and misrepresentation of livability differences across individuals. Using a people-based perspective where indicators are tracked with respect to individuals rather than locations is a useful complement. Both place-based and people-based perspectives are required to capture the full spectrum of livability and its variations.
Human settlement landscapes exhibit substantial and complex variability with respect to time as well as place. Recording livability data for a place only at one particular time can misrepresent urban and regional structure and processes. Livability should be analyzed over time as well as space at time scales varying from daily to weekly, monthly, yearly, and over multiple decades. This should be accomplished using both people-based and placed-based perspectives.
Livability data are often recorded or reported using incompatible spatial units. Appropriate spatial basis transfer methods should be used to integrate these data. The appropriate method depends on
beliefs or assumptions about the spatial variability of the data within the spatial units.
Spatial data can create problems for standard inferential statistical methods that assume observations are independent and the process is uniform across locations. The results from statistical methods that do not consider spatial dependence and heterogeneity are suspect. Spatial statistical methods, such as disaggregate spatial autocorrelation methods and spatial regression analysis, can resolve these problems.
Individuals’ access to activities and resources is important to livability but is often crudely measured. Including space-time constraints in accessibility measures captures the influence of individuals’ activity schedules and major anchor points on their access to resources, opportunities, and activities. These schedules vary substantially by social class, cultural, life cycle, and gender roles. Space-time accessibility measures also can capture the growing impact of telecommunication and information technology on individual accessibility.
Adams, P. 1995. A reconsideration of personal boundaries in space-time. Annals of the Association of American Geographers 85:267-285.
Adams, P. 2000. Application of a CAD-based accessibility model. Pp. 217-239 in D. G. Janelle and D. C. Hodge, eds., Information, Place and Cyberspace: Issues in Accessibility. Berlin: Springer.
Anselin, L. 1988. Spatial Econometric: Methods and Models. Dordrecht, Germany: Kluwer Academic.
Anselin, L. 1993. Discrete space autoregressive models. Pp. 454-469 in M. F. Goodchild, B. O. Parks, and L. T. Steyaert, eds., Environmental Modeling with GIS. New York: Oxford.
Anselin, L. 1995. Local indicators of spatial association—LISA. Geographical Analysis 27:93-115.
Anselin, L., and D. A. Griffith. 1998. Do spatial effects really matter in regression analysis? Papers of the Regional Science Association 65:11-34.
BLM (Bureau of Land Management). 1999. Grand Staircase-Escalante National Monument, Geospatial Data Sharing Pilot Project. Available at http://www.ut.blm.gov/monument/Science_and_Research/papers/fgdc1.pdf. Accessed September 24, 2001.
Brunsdon, C., A. S. Fotheringham, and M. E. Charlton. 1996. Geographically weighted regression: A method for exploring spatial nonstationarity. Geographical Analysis 28:281-298.
Burns, L. D. 1979. Transportation, Temporal and Spatial Components of Accessibility. Lexington, Mass.: Lexington Books.
Dodge, M., and R. Kitchin. 2001. The Atlas of Cyberspace. Boston, Mass.: Addison Wesley Longman.
Flowerdew, R., M. Green, and E. Kehris. 1991. Using areal interpolation methods in geographic information systems. Papers in Regional Science 70:303-315.
Forer, P. 1998 Geometric approaches to the nexus of time, space and microprocess: Implementing a practical model for mundane socio-spatial systems. Pp. 171-190 in M. J. Egenhofer and R. G. Golledge, eds., Spatial and Temporal Reasoning in Geographic Information Systems. New York: Oxford University Press.
Fotheringham, A. S. 1981. Spatial structure and distance-decay parameters. Annals of the Association of American Geographers 71:425-436.
Fotheringham, A. S. 2000. Context-dependent spatial analysis: A role for GIS? Journal of Geographical Systems 2:71-76.
Fotheringham, A. S., and P. A. Rogerson. 1993. GIS and spatial analytical problems. International Journal of Geographical Information Systems 7:3-19.
Fotheringham, A. S., and D. W. S. Wong. 1991. The modifiable areal unit problem in multivariate statistical analysis. Environment and Planning A 23:1025-1044.
Frank, A., J. Raper, and J.-P. Cheylan, eds. 2001. Life and Motion of Socio-economic Units. GISDATA 8. London: Taylor and Francis.
Geertman, S. C. M., and J. R. R. van Eck. 1995. GIS and models of accessibility potential: An application in planning. International Journal of Geographical Information Systems 9:67-80.
Getis, A., and J. K. Ord. 1992. The analysis of spatial association by use of distance statistics. Geographical Analysis 24:189-206.
Golledge, R. G., and R. J. Stimson. 1997. Spatial Behavior: A Geographic Perspective. New York: Guilford.
Goodchild, M. F., and D. G. Janelle. 1984. The city around the clock: Space-time patterns of urban ecological structure. Environment and Planning A 16:807-820.
Goodchild, M.F., and N.-S. Lam. 1980. Areal interpolation: A variant of the traditional spatial problem. Geo-Processing 1:297-312.
Goodchild, M. F., L. Anselin, and U. Deichman. 1993. A framework for the areal interpolation of socioeconomic data. Environment and Planning A 25:383-397.
Griffith, D. A. 1983. The boundary value problem in spatial statistical analysis. Journal of Regional Science 23:377-387.
Griffith, D. A., and C. G. Amrhein, 1983. An evaluation of correction techniques for boundary effects in spatial statistical analysis: Traditional methods. Geographical Analysis 15:352-360.
Handy, S. L. 1994. Highway blues: Nothing a little accessibility can’t cure. Access 5:3-7.
Hansen, W. G. 1959. How accessibility shapes land use. Journal of the American Institute of Planners 25:73-76.
Horn, M. E. T. 1995. Solution techniques for large regional partitioning problems. Geographical Analysis 27:230-248.
Janelle, D. G., B. Klinkenberg, and M. F. Goodchild. 1998. The temporal ordering of urban space and daily activity patterns for population role groups. Geographical Systems 5:117-137.
Jensen, J. R., and D. C. Cowen. 1999. Remote sensing of urban/suburban infrastructure and socioeconomic attributes. Photogrammetric Engineering and Remote Sensing 65:611-622.
Johnson, R. J. 1986. Ecological fallacy. P. 115 in R. J. Johnson, ed., The Dictionary of Human Geography, 2nd edition. Oxford, U.K.: Blackwell.
Kirby, H. R. 1997. Buffon’s needle and the probability of intercepting short-distance trips by multiple screen-line surveys. Geographical Analysis 29:64-71.
Knox, P. L. 1978. Territorial social indicators and area profiles: Some cautionary observations. Town Planning Review 49(1):75-93.
Kwan, M.-P. 1998. Space-time and integral measures of accessibility: A comparative analysis using a point-based framework. Geographical Analysis 30:191-216.
Kwan, M.-P. 2000a. Interactive geovisualization of activity-travel patterns using three-dimensional geographical information systems: A methodological exploration with a large data set. Transportation Research C: Emerging Technologies 8:185-203.
Kwan, M.-P. 2000b. Human extensibility and individual hybrid accessibility in space-time: A multiscale representation using GIS. Pp. 241-256 in D. G. Janelle and D. C. Hodge, eds., Information, Place and Cyberspace: Issues in Accessibility. Berlin: Springer
Kwan, M.-P., and X.-D. Hong. 1998. Network-based constraints-oriented choice set formation using GIS. Geographical Systems 5:139-162.
Martin, R. J. 1987. Some comments on correction techniques for boundary effects and missing value techniques. Geographical Analysis 19:273-282.
Mesev, V., P. Longley, and M. Batty. 1996. RS-GIS: Spatial distributions from remote imagery. Pp. 123-148 in P. Longley and M. Batty, eds., Spatial Analysis: Modeling in a GIS Environment. Cambridge, U.K.: GeoInformation International.
Miller, H. J. 1991. Modeling accessibility using space-time prism concepts within geographical information systems. International Journal of Geographical Information Systems 5:287-301.
Miller, H. J. 1999a. Potential contributions of spatial analysis to geographic information systems for transportation (GIS-T). Geographical Analysis 31:373-399.
Miller, H. J. 1999b. Measuring space-time accessibility benefits within transportation networks: Basic theory and computational methods. Geographical Analysis 31:187-212.
Miller, H. J., and Y.-H. Wu. 2000. GIS software for measuring space-time accessibility in transportation planning and analysis. GeoInformatica 4:141-159.
Monmonier, M. 1996. How to Lie with Maps, 2nd edition. Chicago: University of Chicago Press.
Openshaw, S. 1977. Optimal zoning systems for spatial interaction models. Environment and Planning A 9:169-184.
Openshaw, S. 1978. An empirical study of some zone-design criteria. Environment and Planning A 10:781-794.
Openshaw, S., and L. Rao. 1995. Algorithms for reengineering the 1991 census geography. Environment and Planning A 27:425-446.
Openshaw, S., and P. J. Taylor. 1979. A million or so correlation coefficients: Three experiments on the modifiable areal unit problem. Pp. 127-144 in N. Wrigley, ed., Statistical Applications in the Spatial Sciences. London: Pion.
Ord, J. K., and A. Getis. 1995. Local spatial autocorrelation statistics: Distributional issues and an application. Geographical Analysis 27:286-306.
Rogerson, P.A. 1990. Buffon’s needle and the estimation of migration distances. Mathematical Population Studies 2:229-238.
Scott, L. M. 2000. Evaluating intra-metropolitan accessibility in the information age: Operational issues, objectives and implementation. Pp. 21-45 in D. G. Janelle and D. C. Hodge, eds., Information, Place and Cyberspace: Issues in Accessibility. Berlin: Springer.
Shen, Q. 2000. Transportation, telecommunications and the changing geography of opportunity. Urban Geography 20:334-355; reprinted on pp. 47-72 in D. G. Janelle and D. C. Hodge, eds., Information, Place and Cyberspace: Issues in Accessibility. Berlin: Springer.
Smyth, C. S. 2001. Mining mobile trajectories. In H. J. Miller and J. Han, eds., Geographic Data Mining and Knowledge Discovery . London: Taylor and Frances.
Taylor, P. J., and D. N. Parkes. 1975. A Kantian view of the city: A factorial ecology experiment in space and time. Environment and Planning A 7:671-688.
Theobald, D. 2001. Quantifying urban and rural sprawl using the sprawl index. Paper presented at the 97th annual meeting of the Association of American Geographers, New York, February 27-March 3.
van der Knaap, W. G. M. 1997. Analysis of time-space activity patterns in tourist recreation complexes: A GIS-oriented methodology. Pp. 283-311 in D. F. Ettema and H. J. P. Timmermans, eds., Activity-Based Approaches to Travel Analysis. Oxford, U.K.: Elsevier Science.
Wong, D., and C. Amrhein. 1996. Research on MAUP: Old wine in a new bottle or a real breakthrough? Geographical Systems 3:73-76.
Wong, D. W. S., and A. S. Fotheringham. 1990. Urban systems as examples of bounded chaos: Exploring the relationship between fractal dimension, rank-size and rural-to-urban migration. Geografiska Annaler 72B:89-99.
Wu, Y.-H., and H. J. Miller. Forthcoming. Computational tools for measuring space-time accessibility within transportation networks with dynamic flow. Journal of Transportation and Statistics.