Geoscience Reference
In-Depth Information
Figure 3.2.
Statistical heterogeneity and spatial heterogeneity
Spatial analysis gives, by definition
4
, a central place to space in the analysis of
the phenomena of interest. The simplest method consists of building spatial
descriptors and formalizing that way the spatial structure of the phenomenon being
studied: the most classic example, which is at the basis of numerous spatial models
relies on the hypothesis that distance is a structuring element (distance to the center,
distance to the facilities, distance to the network, etc.). However, we can also
imagine describing a place by a characteristic
and
this same characteristic measured
in its neighborhood: if the example of the average income of the inhabitants of a
municipality is recalled, tests can be carried out in order to measure the effect on a
municipality of average wealth to have wealthy or poor municipalities surrounding
it. Another way of integrating the role of space is to situate the analysis subject in
the meshes of the zonings in which it is located: the academy for a high school, the
region for a city, the climate zone for an agricultural plot. These higher level spatial
entities can be mobilized as a relationship of belonging or through descriptors
characterizing the higher level spatial unit (a rate of school success for the academy,
the GDP for the region, the duration of the episode of drought for the climate zone).
These relationships between zonings are also a form of expression of the effects of
the neighborhood [MAT 07]. In these different cases, space is said to be “active”
because its structure (distance, neighborhood, mesh hierarchy) plays a role in the
analysis and appears to make explicit the differentiations existing in space for
the phenomenon of interest. Therefore, the notion of “situation” is integrated in the
objects' attributes, which, as other attributes, can evolve over time. This type of
statistical table is undoubtedly the one which has been the most studied in the field
of geographical data analysis.
4 In this context, Charre's definition of spatial analysis is particularly eloquent: spatial
analysis consists of “searching for, in the characters of spatial units, what refers to their
geographical position, in particular relative, which makes it necessary to model the spatial
structure” [CHA 95].
