Geography Reference
In-Depth Information
Multivariate Data Visualization
and Usability: Preliminary Notes
Petr Kub´
ˇ
ek, Radka B
´ˇ
ov
´
, and Zdenˇk Stacho
ˇ
Introduction
Effective and understandable communication of geographic information is cur-
rently under discussion by the scientific community in general and cartographers
in particular. New data sources as well new visualization methods are re-opening
the discussion about the user
s ability to read the map and interpret its content in a
'
coherent way.
Multivariate mapping techniques used for representation of three or more vari-
ables are even more challenging for individuals with a limited cartographic back-
ground. Several types of visualization techniques exist for representing multivariate
spatial patterns on the same map. These include trivariate choropleth maps (Brewer
1994
), multivariate dot maps (Rogers and Groop
1981
), or multivariate point
symbol maps (glyphs) (Chernoff and Rizvi
1975
; Dorling
1994
). Another possibil-
ity is to compare three or more maps at the same time. Such visualization is termed
small multiples (Tufte
1990
; Kousoulakou and Kraak
1992
; Slocum et al.
2009
). A
more complex discussion about multivariate visualization techniques can be found
in Guo et al. (
2005
) or Slocum et al (
2009
).
With respect to the data that can be represented, several authors have found that
multivariate depictions can be used successfully for representation of both temporal
and non temporal data with a wide variety of data nature (nominal, ordinal, interval,
and ratio). However, only preliminary discussion with users has been held regard-
ing their ability to use such representations. In this empirical study, it was decided
to compare the effectiveness and efficiency of two multivariate mapping methods—
ring maps and small multiples—for both spatial and temporal data. The choice of
small multiples is justified by the following points. Small multiples are traditionally
