IMAT 3712

Human Computer
Interaction
Data Graphics
Acknowledgements
 Some slides originated in lectures developed by
Howell Istance.
 Examples of data graphics taken from Edward R.
Tufte’s books and website
 Several data graphics shown here were designed
by Nigel Holmes
Information visualization
Making information visible – so the user can perceive
what it means directly, without need to reason
about it

 Comparisons
 Trends over time
 Degree of correlation between variables
 Outliers

 Effective information presentation can change the

world
Perceptualization
Computer graphics are used to show information

 The aim is to convert a

 cognitive load (understanding requires thinking)

into a
 perceptual load (understanding requires

seeing)

 We call this perceptualization.

 This means that instead of the user having to work
something out or make comparisons in his or her
head, differences in direct sensory input are used
to distinguish between different situations or
Data Visualization
Infographics vs

Illustrations from Killerinfographics.com

Data graphs
Many data graphs in newspapers, etc, are very poor,
often because their designers don’t take them
seriously

Aim to do better!

 Don’t regard as mere decoration

 Consider users’ information needs
 Avoid distortions: area values, 3D effects
 Consider salience effects

See E.R. Tufte, The Visual Display of Quantitative Information,

Graphics Press, 1983.
 PowerPoin
t abuse

© De Montfort University 2012

What does this mean?

© De Montfort University 2012

Visualisations: Design issues
 Find a mapping between the model and the display
which makes perceptually prominent the features
which are conceptually prominent.

 Make the mapping in a principled, consistent way

rather than in an ad hoc fashion.
William Playfair in 1801:
Per-capita taxation

© De Montfort University 2012

Too much tax?
 William Playfair wished
to stress that for the 4
great empires of his
day, the British Empire
was too heavily taxed
 Graphic to make a
point
 Left-hand uprights α
gross national product  Is slope the best way of
 Right-hand uprights α drawing attention to
tax income odd one out?
 Slope α per-capita tax
Florence Nightingale,
heroine of information design
 As a result of her observations of
conditions in the Crimea in the
mid-1850s, Florence Nightingale
persuaded a Sanitary
Commission to undertake
improvements
 She wrote a report describing
the unsatisfactory conditions
and how they were improved.
 The improvements were

displayed in a polar diagram

(‘Bat’s Wing’)
Death rates in the Crimean War
 In this diagram,
Nightingale resolved
the problem of the
"bat's wing" by using
areas to represent the
variation in the death
rate.
 This figure makes it
abundantly clear that
far more deaths were
attributable to non-
battle causes
("preventable causes")
than to battle-related
Death rates in the Crimean War
 Death rate α length of radial line
 BUT appearance implies that it is the shaded area
that is proportional to the death rate
Dr Snow’s cholera map
 Dr John Snow was medical officer for London
tasked with bringing a Soho cholera outbreak
under control in 1845

 He plotted individual deaths on a

map (black dots) and the positions
of water pumps (marked x)
 Noticed that most deaths were
concentrated around the Broad
Street pump
Big picture AND detail:
Buying shares in Enron

© De Montfort University 2012

War of 1812
Macro versus micro views
 "If we are going to make a mark, it may as well be
a meaningful one" - John Tukey

 When did American units

arrive in France in WW1?
 Shape conveys quantity
 Numbers name units
Macro vs micro views
 Heights of volcanoes
 Shape shows distributions
 Each digit gives more information about 1

mountain
Small multiples for effective
comparison
 Chinese pollution
 Several possible comparisons
 Not too many to swamp the reader
© De Montfort University 2012
Signalling information
Information carried by
 Colour
 Shape
 Size
 Font etc
 Orientation
 Alignment
 Spacing

Use or keep neutral!

Choices
 Qualities of
points lines and
areas should be
deliberately
chosen
Tufte's design principles
 Eliminate every pixel not serving user important
data
 Mute minor visual elements
 Remove distracting patterns
 End clutter
Remove distracting patterns!
 Brazilian textile production in 1927…
Box plot
 One glyph per group of
data points
 Shows values of data
points in group by rank
in distribution
 Median data point –
bar in middle
 Data points at 75% &
25% – top and bottom
of box
 Data points at 90% &
10% - outer bars
 Outliers - circles
Better than a box plot?
 Tufte's version is cleaner
 Box plots have advantages:
 Standard notational convention
 Distinctive
 Stress middle

half
Extrasolar planets
 This is now seriously out of date…
 Note the logarithmic scales on both axes
Which display tells the story?
 S.S. Stevens’ research on psychophysics
 Subjective experience of magnitude is a power

function of objective size of stimulus

 Coefficient depends on what stimulus is
 Straight line on a log-log plot
Cancer survival (published)

© De Montfort University 2012

Cancer survival (improved by
Tufte)

© De Montfort University 2012

 Cancer
survival
(graphed
by Tufte)

© De Montfort University 2012

Baseball win-rates

© De Montfort University 2012

Showing numbers: Bar Charts
 Purposes matter for choosing charts…
 Lengths show sizes clearly
 IF you don’t distort them…

 Simple Bar Charts

 GOOD for
 Showing sizes clearly
 Sequential comparisons
Showing numbers: Pie Charts
 Purposes matter for choosing charts…
 Angles are difficult to judge

 Pie Charts are

 GOOD for
 Rough idea of proportions
 BAD for
 Comparisons of similar quantities
Chart Junk:
Age of college students
 Is this the worst data graphic
ever published?
 5 data points
 Spurious curves for discrete data
 Below and above age 25 curves

aren't actually mirror images

Chart Junk?
 Extraneous stuff can obscure message
 Entertaining images make message more
memorable

US election
campaign costs,
as drawn by
Nigel Holmes
Chart Junk?
 Okay, you saw that US House and Senate
campaign costs increased between 1972 and
1982…
 But by how much?
 And what about inflation?
Chart Junk?
 Actual increase: 1982 expenditure 4x larger than in
1972
 You can see this, if you look!
Chart Junk?
 Another Nigel Holmes image… it's more
memorable
 BUT do you take away the right message?

© De Montfort University 2012

Lying with data graphics
 Are sizes shown without graphical distortion?
 Are sizes shown without perceptual distortion?

 Scale from zero, or not?

 Is log-linear scale appropriate here?

 Is it clear enough?
Fox News: Fair and balanced
 A classic use of arbitrary non-zero baseline for
propaganda (US tax rates for richest)
Perceptual distortion
 Perceived length α actual length
 Perceived area α actual area0.7
Lies, damned lies and statistics
 Beware 3D effects…
 Looks like downward trend
 Not obvious that numbers for 1995 and 1997 about
equal
Lies, Damned Lies and
Statistics

This graph purports to show the mandated fuel

economy standards set by the US Department of
Transportation. The standard required an increase
in mileage from 18 to 27.5, an increase of 53%.
The increase shown in the graph is actually 783%
Lying with data graphics
 Easy to do…
 Easy to do just for the sake of decoration
 Take your professional ethics seriously!

 Does the chart show the true story?

 Does the chart show the true story without
distortion?
Remember…
“Clutter and confusion are failures of design,
not attributes of information”

 Edward R. Tufte
Reading
 Chiqui Esteban, A quick guide to spotting data
graphics that lie, National Geographic
 http://news.nationalgeographic.com/2015/06/150619-da
ta-points-five-ways-to-lie-with-charts/
 Walter Hickey, EXPOSED: Here Are The Tricks
That Fox News Uses To Manipulate Statistics On
Its Graphics, Business Insider
 http://www.businessinsider.com/fox-news-charts-tricks-d
ata-2012-11?IR=T
Some good books
How to design graphic displays of information

 Harris, R.L. (1999) Information Graphics: A

Comprehensive Illustrated Reference, Oxford
University Press
 Tufte, E.R. (1983) The Visual Display of
Quantitative Information, Graphics Press.
 Tufte, E.R. (1990) Envisioning Information, Graphics
Press.
 Tufte, E.R. (1997) Visual Explanations, Graphics
Press
Another good book
How to do graphic design for screen displays

 K. Mullet & D. Sano, Designing Visual Interfaces,

SunSoft/Prentice-Hall, 1995