Partitioning of Data - Spatial Mathematics: Theory and Practice through Mapping - page 132

Figure 6.6 Attribute table of maps in Figures 6.1 through 6.5 graphed in Microsoft
Excel.
What is critical to understand here is the importance of considering how one
classifies data sets in advance of presenting a final map. When the final map is
presented, it is also necessary for the mapmaker to explain decisions that were
made in regard to partitioning of data because the results can vary greatly
depending on these fundamental decisions. Equally, it is critical for the map
reader to ask questions about how data sets were partitioned in advance of
accepting apparent spatial pattern presented on maps derived from underly-
ing data sets—it is far too easy to gerrymander results—we should take to
heart Monmonier's (1996) cautions in interpreting maps!
6.3 Normalizing data
In the previous section, we illustrated the importance of choosing an appro-
priate ranging technique to display data—bad choices lead to bad results. Of
equal, but different, importance is the issue of normalizing data. The maps
of the previous section were all based on total land area by country. Often,
mapping totals might be an informative way to discover patterns in the data,
in this case, mapping how many square kilometers each country occupies.
With other types of data, especially with population data, merely mapping
total counts may be insufficient.
Consider Figures 6.7 and 6.8. In Figure 6.7 , the population of the United
States is mapped by county and displayed as total counts by county. The
classification chosen was Quantile in five ranges. In the region of California,
Nevada, Utah, and Arizona, many of the counties appear quite large in land
area; and by virtue of having a large land area, many of these counties also