RECENT
IMMIGRANTS AND UNDERCLASS POPULATION IN CANADIAN CITIES
SPATIAL
ANALYSIS
Modifiable Unit
Area Problem (MAUP)
Benjamin Davis (2003) observed that: A major problem in spatial analysis is
modifiable area unit problem (MAUP). Units of area, whether
administrative or political boundaries, agro-ecological zones or image
pixels, are essentially arbitrary groupings, and the data within can be
aggregated in an infinite number of ways (Nelson, 2001; Bigman and
Deichmann, 2000a). This includes GIS-constructed data and any kind of
spatially aggregated data such as censuses or household surveys. The
practical implication is that alternative aggregations of the data may
lead to different and conflicting results. This is true for simple
visual correlation analysis as well as sophisticated econometric
techniques. In terms of multivariate analysis, the relationship between
dependent and independent variables may change over space in a manner
that the analyst may not be able to determine a priori. Results can
thus be purposely modified, or errors inadvertently made, through the
process of aggregating data.
Group of interrelated problems
associated with MAUP are:
Scale problem: tendency for different statistical results to be
obtained from the same set of data when that information is grouped at
different levels of spatial resolution.
Aggregation problem: variability in results obtained through
variations in the shape of areas.
Ecological fallacy: when statistics measured on spatially
aggregated data are used as a substitute for statistics based upon
individuals.
The maps developed in figure….,
analysed the aggregation problem in relation to census tract with
enumeration area in 1996 and census tract with dissemination area in
2001. Spatial distribution changes with variation in shape of
aggregation area. Aggregation at a large area such census tract assigns
a greater spatial coverage to the same attribute that might be
concentrated at one zone within coverage area. This may result in
over-representation for small spatial areas with disperse attribute
value variability or under-representation for large areas with
homogeneous spatial variability in attribute value. Consequently,
spatial definition for attribute values for large area aggregation
becomes poor because the aggregation approach pre-defined both the
spatial accuracy and the spatial statistical analysis made therein. In
this fashion, all analytical techniques employed on the data are
subject to the spatial resolution of the aggregation area.
Solution to MAUP
Spatially disaggregated data:
Disaggregated data is data representation by test scores of specific
subgroups of individuals. The use of aggregated data to explain
individual behavior relies on the assumption that the hypothesized
relationship between the spatial variable in question is homogenous
across all individuals. It is not, however, until the data are
disaggregated that patterns, trends and other important information are
uncovered. However, such data are not available for reasons of
confidentiality (Fotheringham A. S., et al 2000).
Report results of spatial
analysis at the most disaggregated level possible and then demonstrate
visually the sensitivity of the results to both the scale and zoning
effects. In this way, if some results can be shown to be
relatively stable over a wide range of zoning systems, this can induce
greater confidence that the results at the most disaggregated level
have some meaning and are not simply artifacts of the way the data are
arranged(Fotheringham A. S., et al 2000).
Optimal zoning systems: For
Openshaw and Rao (1995) the answer lies in the construction of zoning
systems that are in some sense optimal. It is not always clear whether
this construction is intended to influence those releasing the data. It
might be regarded as somewhat eccentric to create a zoning system for
which the fit of a particular spatial model is optimal if the zoning
system is to be for some completely different purpose (Fotheringham A.
S., et al 2000).
Two different sizes of
aggregation - aggregation problem
Two
different sizes of aggregation - aggregation problem
