Geography Reference
In-Depth Information
Fig. 8 Crime type mapped to degree grid squares covering Brazil (as % for each grid square):
a Robbery; b Violence; c Theft; d Attempted robbery; e Attempted theft
(Fig.
7
v). The top five reasons had large majority coverage. Unemployment and
gunman, along with the 12 specific reasons not mentioned above, had about two-
thirds of states covered (down to just 5 states out of 27 covered for police violence).
3.1.4 Choropleth Maps By Degree Grid Cell
Figures
8
,
9
,
10
,
11
use a regular one degree grid of polygons covering the
Brazilian land area as the geometric basis for the choropleth maps (a grid
choropleth—Kraak and Ormeling
2010
). This is to address the widely acknowl-
edged ecological fallacy that occurs (Robinson
1950
). In this case, an entire state
area is assumed to adopt a single attribute value when in fact that value masks a lot
of variation in attribute magnitude within that area. Although a degree grid cell
suffers from the same effect, it is lessened, and at the Brazil scale, permissible.
The irregular state boundaries also denote an extent that affects the aggregated
data values in a misleading manner. Alter the boundaries even slightly and you
may derive a very different aggregate value—due to ecological fallacy, most of the
point occurrences affecting an aggregate value attributed to an area may occur near
the boundary. If the boundary shifts, some or most of the points may be left outside
the area, implying a change for the aggregated value. This is the well-known
phenomenon of the Modifiable Areal Unit (MAUP—Openshaw and Taylor
1981
),
the effects of which are minimized by applying a regular grid with an adequate
resolution for the scale of a large country.
