Geography Reference
In-Depth Information
fixations duration observed at specified AOIs
was analysed with Kruskal-Wallis test. Pairwise comparison of data related to the
colour distance and the class order was done in order to find out statistically
significant differences of median location. Analyses were done separately for
each AOI. In the case of equally-stepped schemes only imperceptible number of
compared pairs was found significantly different, thus only elementary visual
interpretation of results can be done based on boxplots (Fig.
4
).
The sum of fixations spent in AOI 1 is generally the highest among other AOIs.
Higher medial values of fixations duration can be observed at classes B-F, which
correspond to the previous findings. Looking at boxplots for AOI 2 (close region to
the depicted area) it is clear that lightest and darkest classes (A and F) in most cases
shows the smallest, even zero, duration of fixations. Observations for AOI 3 are
more or less balanced, which means that participants spend the same (obviously
short) time in this AOI regardless the colour distance or class. Values at AOI
corresponding to the legend (AOI 4) are tracing values of AOI 1; even observed
fixations durations are much shorter.
The same analysis was done for the
Dataset containing all participants
'
optimised colour schemes (4-8-
10-8-4). Kruskal-Wallis test did not prove any significant differences as well as for
schemes with equal colour distance step. However in the case of all AOIs the
medial values are by visual control more equable (see Fig.
5
), than in original
colour schemes.
winning
'
'
Conclusion and Discussion
Results of the experiment didn
t prove any influence of legend position on the
map sheet on the effectiveness of the map reading process. The effect of the
colour distance on the accuracy of reading the choropleth maps was con-
firmed. The higher is the colour distance between classes of the colour
scheme the more accurate are map-users while estimating the value of
concrete area comparing it with the map legend. Another interesting finding
is that equal colour distance steps are not optimal solution in designing
sequential schemes. Map users perform particularly better when determining
minimal and maximal values, which are plotted by the lightest, respectively
darkest colour shade. Based on these findings new colour schemes of unequal
visual steps between classes were designed and evaluated. The most efficient
colour scheme was the one with colour distance steps 4-8-10-8-4. Observed
accuracy was for this particular scale balanced regardless the class order and
the overall rate of correct answers achieved 93 %.
In the presented experiment only lightness continuum of six green shades
was evaluated. It is known that the human eye maximal sensitivity is in the
green region at 555 nm. Even though in theory the human eye can distinguish
millions of colours, in cartography would be unreasonable to ask map users to
distinguish more than 12 classes (Dent et al.
2009
). It is assumed, that the
'
(continued)
