Geography Reference
In-Depth Information
et al. 2006 ) and for mobile applications (van Elzakker et al. 2008 ). Various papers
can be mentioned as examples: Sedl ´ k et al. ( 2011 ) focused on evaluation of map
portals, ˇ t ˇ rba et al. ( 2011 ) focused on map symbols, Popelka and Brychtov ´
( 2013 ) dealt with 2D and 3D visualizations, and Fabrikant et al. ( 2008 ) focused
on small multiples and eye tracking. Rogers and Groop ( 1981 ) made one of the first
empirical evaluations in the field of multivariate data visualization. They compared
the multivariate dot map with several univariate dot map methods and evaluated
map reading tasks. Rogers and Groop proposed that the multivariate method was
more effective for perceptual tasks, but also pointed out the limited legibility of the
multivariate method for more than three categories.
Kaspar et al. ( 2011 ) performed a study comparing contiguous cartograms and
choropleth maps for spatial inference making. They confirmed significant differ-
ences in the user
s performance according to the visualization type and the domi-
nance of choropleth maps over cartograms. Usability of the visualizations was also
dependant on the complexity of the map use tasks and on the shape of the
enumeration units. van den Elzen and van Wijk ( 2013 ) compared small multiples
and large singles for both the effectiveness and efficiency of visual exploration.
They discovered no advantages of small multiples in efficiency (execution time) or
in the number of errors (correctness of answers). However, they found that users
were more satisfied and preferred exploration methods using small multiples. All
the studies generally confirmed that overall efficiency and effectiveness depends
not only on the visualization method itself, but also on the complexity of the tasks
(map reading, information exploration, and information inference) and the visual
variables used.
'
Data Used and Preparation of the Test
To empirically evaluate the difference between particular types of multivariate
representation, we designed and performed an empirical experiment. Two different
types of visualization—namely ring maps and small multiples—were used for three
types of tasks.
Health data, which are typical representatives of large and multivariate data sets,
were used for map preparation. We created two different base maps for the
empirical experiment using two cartographic visualization techniques—attribute,
and temporal characteristics.
The first base map (attribute) contains the following data:
• Main characteristic—Number of heart attack deaths per 10,000 people—(shown
on a central choropleth map in ring maps, or on one of the choropleth maps in
small multiples) was classified into five categories.
• Secondary characteristics—Crude mortality rate, Percent of urban population,
GDP per capita, Unemployment rate, and Pharmacy profits from supplements in
thousands of CZK (shown on rings in ring maps, or on the remaining choropleth
Search WWH ::




Custom Search