Making Better Graphs

1
MakingBetterGraphs
Mike Dickison www.numberpix.com www.giantflightles
sbirds.com
2
Two ways to make bad graphics Cruddy versus slick.
3
Why are academics suspicious of well-designed
graphics?

• Slick glossy marketing brochure ? lies, lies,
  lies
• Crude old-school luddites ? Nobel laureates
• Appearance good ? content bad ? Ab ? Cg (?)
• Nice graphics mean you have too much time on your
  hands. Get back to work!
• Fancy graphics are spat out by some fancy
  graphics package (and thus untrustworthy).
• Fancy graphics are crafted by some fancy designer
  (and thus untrustworthy).

4
Why data graphics?A good table is better than a
bad graphic.
5
The OctantsEight solutions to anygraphics
problem.
6
Show the data if possible(and its usually
possible)
7
Make the furniture recede(maximizing the
dataink ratio)
8
Make the furniture recede(maximizing the
dataink ratio)
9
Avoid unnecessary colorColor costs money, and
disappears on printing/faxing/copying. Use it for
optional emphasis.
The Scientists Rainbow
10
Use an intuitive color scaleColor should work
with the data to support meaning, not fight or
obscure it.
11
Use an intuitive color scaleColor should work
with the data to support meaning, not fight or
obscure it.
12
(No Transcript)
13
(No Transcript)
14
Label directlyPlace labels directly on the graph
rather than in a key
15
Workflow
Redraw
16
Why make axes hard to read?Use real units, real
scales, sensible labels.
200
400
600
800
1000
0
Elapsed post-dispersal (d)
17
(No Transcript)
18
Contrast in heading, subheading, and labels Same
font, large axis labels, small graph title No
color gray black Consistent color and
fonts Bars overlapping Direct labelling
Small as possible, but plenty of data No frames,
no background colors or decoration Bars have a
fill, no stroke No y axis line Lightly dotted y
gridlines Years abbreviated
No tickmarks, but dividers between
years Unobtrusive gaps used between bars. Can see
shape of series. Explanations on the chart Color
scheme intuitive (general gt specific gray gt
black) Ohio series thinner, only when it needs to
be (for overlap)
19
Label directlyIf possible, make each data point
its own label
20
Label directlyIf possible, make each data point
its own label
21
(No Transcript)
22
(No Transcript)
23
Small multiplesto show variation in time and
space
Whats the intuitive direction here?
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29

• Why graphics at all?
• The octants audience, explain/explore,
  qual/quan.
• Whats the story?
• Make the furniture recede maximize dataink
• Many ways to convey difference
• Make axes as readable as possible
• Show the data if you can
• Use color only when necessary
• Choose an intuitive color scale
• Label directly, making the data its own label
• Use small multiples to show time, variation

30
Further Reading
Edward Tufte The Visual Display of Quantitative
Information (and others) William Cleveland The
Elements of Graphing Data,Visualizing Data Robin
Williams design books www.giantflightlessbir
ds.com/workshops (handouts, this
powerpoint) www.numberpix.com (info design blog)