Geoscience Reference
In-Depth Information
where z
x
and z
y
are the z-scores (distances from the mean expressed as
units of standard deviation) associated with x and y respectively.
Spearman's rank correlation coefficient ( ρ ) provides a measure of the lin-
ear association between the two ranked variables which varies between -1
(perfect negative correlation) and +1 (perfect positive correlation) (
ibid
.,
p.150). Each set of scores is ranked separately from lowest to highest and a
Pearson's correlation test (as above) is then calculated on the ranks, mak-
ing the technique suitable for testing the degree of association between two
variables that may not correlate linearly (Hinton 2004, p.279, 280). Hence,
where
d
is the difference between the ranks for observation and
n
Is the size of the sample.
Regarding the quantitative data derived from the maps, comparing the val-
ues for each symbol type as a percentage of the total symbology can be
more useful than using symbol counts because this gives some indication
of the relative meaning associated with each type of feature. While per-
centage values will be incorporated into these tests, what can offer more
significance are those correlations between the raw symbol counts and
various development statistics. This provides the added dimension of the
relative magnitude of symbologies within the sample. Therefore, both
count and percentage data are utilized in the following analyses.
3- Results
3.1 Legend Languages
Before considering factors such as the size of a country, its population
density, and employment structure in order to determine the extent to
which topographic maps reflect these conditions, the first criterion to be
examined is the languages employed in the legend explanations.
Table 2
presents a summary of the mother-tongue languages spoken in the sample
