Is the medium the message

1
Is the medium the message??

• Suraje Dessai gives a nice overview of the rather
  under-developed status of the graphic uncertainty
  communication field, and he rightfully stresses
  the importance of Context in relation to
  Communication Interpretation
• I have just a few remarks to add
• Surajes focus is on quantifiable (mostly
  statistical) uncertainties. What about graphical
  display of qualitative uncertainties (e.g.
  pedigree scores)?
• What about displaying uncertainties in maps
  (spatial information)?

2
Displaying Qualitative uncertainty information

• E.g. info on pedigree scores
• Problems
• How to give a decent assessment of pedigree of
  the information in a graph (e.g. the results of
  model A can have different pedigree than those of
  model B etc., what if results of model A and B
  are displayed in the same graph?)
• Beware of misleading visual artefacts when
  displaying a pedigree in graphical form, e.g.
  kite diagram (see next sheet).

3
Looks nice, but the visual look can be very
illusive (e.g. a similar situation with two high
scores (3) and two low (0) give completely
different display on different axes) -gt improve
by e.g. using equally sized block diagrams,
instead of axes.
4
Displaying uncertainties in maps

• What grid-size choice is warranted, given the
  uncertainties in the data ? (e.g. landuse in
  Europe in A2-scenario in 2030)

5
Displaying uncertainties in maps

• What classification to choose in legenda when
  displaying information (regardless whether
  uncertain or not) --gt different patterns (see
  below)
• see also Tufte, Visual Explanations
• or www.personal.psu.edu/users/k/m/kmb344/lab5.htm

6
Displaying uncertainty in maps

• In case of uncertainty Choosing the
  classification in the legenda is an important
  issue the error of misclassification can be
  different for the classes
• ----gt see example below which illustrates some
  of the characteristics e.g. when the relative
  uncertainty is e.g. ? 10 then the error of
  misclassification grows on a linear scale,
  consider e.g. equidistant classes 0-1, 1-2, and
  2-3, 3-4 etc..

7
Much remains to be explored see also (a
small, aselect sample from the Internet)
http//www.cse.ucsc.edu/research/avis/unvis.html
http//www.gstat.org/bin/apm.pdf.
http//www.colorado.edu/research/cires/banff/pubp
apers/115/