Clear and To The Point. Stephen M Kosslyn
Clear and To The Point. Stephen M Kosslyn
Clear and To The Point. Stephen M Kosslyn
Stephen M. Kosslyn
1
2007
1
Oxford University Press, Inc., publishes works that further
Oxford Universitys objective of excellence
in research, scholarship, and education.
Oxford New York
Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto
With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam
To my wife, Robin
Contents
Preface
ix
Chapter 1.
Chapter 2.
19
Chapter 3.
Legible Text
61
Chapter 4.
99
Chapter 5.
125
Chapter 6.
163
Chapter 7.
191
Appendix:
201
Index
215
Preface
Preface
trouble, and good sense were extremely valued at every step along
the way. Also at Oxford, Lelia Mander did a superb job in managing
the production of this book, and I again thank her for her clearheaded, good judgment. My agent, Rafe Sagalyn, not only read the
manuscript and made extraordinarily constructive suggestions, but
he also broke the logjam over the title (and came up with the one
that graces the cover). Rafe is everything an agent should be and
much more. I also thank my assistant, Alex Russell, for lending me
her sharp eye, and my student Sam Moulton, PowerPoint maven
supreme, for going over the manuscript in detail and sparing no
words in setting me straight. (Any remaining errors are not his fault,
but rather reflect my own foibles.) Andrew Shultman and Patricia
Leigh Zadnik also helped make the earliest drafts begin to approach
coherence. And Nicholas C. Liu helped put the manuscript together,
with patience and good humor, Jennifer Shephard helped in too
many ways to enumerate, and Stefano Imbert and David Kosslyn
provided invaluable image design ideas and produced some of the
most interesting illustrations in the book. Ronnie Lipton gave the
manuscript an extraordinarily close read, and made invaluable suggestions (as well as uncounted specific edits). Thanks, Ronnie!
Finally, I must thank my family: Robinfor giving me the space to
do this project, in spite of looming deadlines on our jointly authored
textbooksand our children, Justin, David, and Neilall of whom
contributed valuable insights and observations.
Introduction to Articulate
Presentations
hatever
business
were
in,
Objects that are near each other: We see XXX XXX as two groups
of 3 Xs, whereas we see XX XX XXthe same number of X
marks, differently arrangedas three groups of 2 Xs.
Objects that appear to be similar: We see XXXOOO as two groups.
Objects that line up on a smooth curve. We see as a
single line, but as two lines (each of which is a group).
Objects that move similarly: We see two words flying in from the
left side of the display as grouped together.
Objects that form simple shapes. We see [ _ ] as a single group, but
not _ ] [
Example: Labels should be closer to the object (or, in a graph, to
the wedge, line, or bar) that they label than to anything else.
10
11
12
God Does
Not Exist
God Exists
Believe
Dont
Believe
Figure 1.1. A summary of Pascals Wager, which allows the audience easily to
understand and retain the structure of the argument for believing in God.
In the remainder of the book, I will offer many specific recommendations based on these eight principles, some sweeping and some
that may seem nit-pickingbut all will help you to design and deliver
compelling presentations. The principles apply at every level of scale,
from the presentation as a whole all the way down to individual illustrations on a single slide and the design of a single bullet point.
13
Figure 1.2. Minards depiction of Napoleons 1812 Russian campaign (from Tufte, 1983). See also color version, in insert.
15
16
17
18
20
1.
2.
3.
21
22
4.
Each slide should contain only as much as you can read aloud or
describe in about one minute. If you present too much text or too
many graphics, the audience will be looking at one thing while you
are saying another.
Vary the salience of what you show by making whats important
perceptually distinct. Capture the audiences attention by making
important elements larger, brighter, or louder, so that you control
what the audience members pay attention to.
How you highlight your material should depend on the type of
material you are highlighting: If you want the viewers to understand a complex structure (e.g., an organizational chart), it makes
sense to build up a slide one part at a time, only showing the part
you want to talk about at that moment. If you simply want them to
focus on a specific part, it makes sense to build a pointer into
your slide by including a red arrow that points to the subject of
interest, or put a circle around a portion of a graphic or text that is
your focus.
5.
6.
23
24
7.
8.
1.
Your very first slide should define the topic and set the stage for
your presentation. To do so, it can simply provide a title, your name,
your affiliation, and any other vital information. (For details about
titles, see Chapter 3.)
One way to construct an effective opening has two distinct phases:
25
26
2.
Figure 2.1. A title slide, which conveys the key information about the topic
and presenter. Please note that the illustrations in this book can only approximate how slides will look when projected by a particular projector. So there
is no substitute for actually viewing a presentation as it will appear to the audience before delivering it, and adjusting contrast and color as necessary.
Figure 2.2. An opening slide that engages the audience. Although this is a work
of art, one can ask how it would function to communicate a specific message
and point out some of the problems with using it in this way. (Les Papillons
dAnna, reprinted with permission from Kss. See color version in insert.)
28
3.
Using Psychological
Principles to Create
Clear and Compelling
Graphs
Figure 2.3. This slide introduces the topic of the presentation. The black frame
helps to draw the viewers attention to the text.
Figure 2.4. This slide acknowledges that many people have an overly narrow
view of what psychology is and believe that it pertains only to psychotherapy.
The audience needs to be disabused of this idea early in the presentation. (See
also color version, in insert.)
Psychological Principles :
The nature of mental
processes, such as those
used in :
Perception
Memory
Cognition
Figure 2.5. After saying what psychology is not (i.e., entirely focused on
psychotherapy), I need to add some of the things that it is; this slide prompts
me to explain what I mean by the term psychological principle.
29
30
4.
Figure 2.7. This graph presents the same information shown in Figure 2.6, but
the atypical group now is obvious at first glance.
just a script for what you will say aloud. That will pique the audiences
interest, that will make them wonder how you will get from here to
there.
Example: In my presentation, I prepared Figure 2.8 to demonstrate that graphs arent better than tables in general. I point out
that Figure 2.8 displays the identical data as Figure 2.7 but plotted
differently; now its not immediately clear which group deviates
from the others. Thenas planned in advanceI ask, Why not?
This rhetorical question lets me tell the audience where I am
going in the presentation. (The answer, by the way, is that our
visual systems are not good at detecting differences in differences of heightand thats how the relevant information is
conveyed in this particular graph. We humans are terrific at
comparing differences in tilt, but not differences in differences
of heights. But for all we know a Martian, with a different brain,
might have the opposite strength and weakness.)
In setting up this sequence of slides, Ive looked ahead to a key
part of what I want the audience to conclude: Namely, that our
ease in understanding the message in Figure 2.7 versus our difficulty with exactly the same information in Figure 2.8 reflects
fundamental facts about how our perceptual systems work, not
attributes of the figures or the data themselves.
31
32
Figure 2.8. This graph presents the same information provided in Figures 2.6
and 2.7, but the atypical group is no longer obvious at first glance. The point:
Its not that graphs are always better than tables; graphs are superior only
when they play to our perceptual strengthsand avoid relying on our weaknesses. We are good at noticing differences in tilt (Figure 2.7), but not differences among differences in height (this graph).
5.
Three Goals
Figure 2.9. An outline of the three goals that will be addressed in the presentation; I briefly explain each one.
33
34
1.
Tell a story.
The human brain automatically tries to organize and make sense out
of experience. One happy consequence of this tendency may be that
humans like to tell and understand stories. Take advantage of this
proclivity by telling a story in the body of your presentation. In order
to do so:
2.
Create a clear line of argument from the beginning to the end of the
body, along which you build a case for the conclusion or conclusions you want to draw.
Ensure that the transitions between parts are clear.
Organize the parts so that each provides the foundation for the
next.
3.
Three Goals
Connect and Engage
Direct Attention
Promote Understanding
and Memory
Figure 2.10. The same outline as in Figure 2.9, but with only the top entry
salient. By graying out the other entries, I direct the audience members attention to the relevant heading, which introduces the following section of the
presentation.
35
36
Figure 2.11. Because the following part is composed of two or more subparts,
I present an outline of the subparts. As shown here, the names of the subparts
should look distinctive and less salient than the name of the part itself.
4.
5.
Figure 2.12. I introduce each subpart by graying out the others, leaving salient
only the name of the subpart about to be discussed.
6.
37
38
Principle of Relevance:
Communication is most
effective when neither
too much nor too little
information is presented.
Figure 2.13. It is important to define key terms in writing (remember: use the
slides as your notes), and read the definition aloud. This gives the audience
two chances (visual and verbal) to absorb this crucial information.
Figure 2.14. If this slide were used to convey specific data, it would provide
too much information for the audience to absorb. To drive home this point,
you could ask the audience to try to imagine in what situations so much information in such a complex display would be relevant. (Reprinted with permission by Scientific Programming Enterprises.)
[Don't]
Figure 2.15. This graph illustrates a more common type of display that
contains more information than can be absorbed easily. If it is intended
to make a particular point, not all of this information is likely to be
relevant.
7.
8.
39
40
[Do]
Figure 2.16. This graph presents only the information in Figure 2.15 that is
relevant for the discussion, sparing the viewers from having to search for this
information.
Three Goals
Achieving additional
goals is facilitated by
additional principles,
used in combination
with the earlier ones.
Figure 2.17. Before moving to the material in the next major part, I present a
transition slide to remind the audience that principles in the service of the first
goal also help to accomplish the following goals.
Promote Understanding
and Memory
Compatibility
Informative Changes
Capacity Limitations
Figure 2.18. This slide shows the outline of the three subparts for the final
major part of the presentation.
41
42
9.
Prepare demonstrations.
As you saw earlier, people automatically pay attention to things that
are clearly different. And there is nothing more different than being
asked suddenly to shift from being a passive listener to being an
active participant. Plus, as you now know, actively processing information helps people understand and remember.
To keep your audience alert and engaged, prepare a demonstration or even simply ask them to raise their hands if they agree (or
disagree) with a specific point.
Example: In Figures 2.22 to 2.26 I have a demonstration of the
Principle of Capacity Limitations, focusing on short-term memory
limits. Heres how I set it up: When I present Figure 2.22, I read the
slide aloud, telling the audience that Ill read a set of directions,
such as the ones they are about to see. Then I show Figure 2.23,
Promote Understanding
and Memory
Compatibility
Informative Changes
Capacity Limitations
Figure 2.19. After presenting the material for each subpart, I revisited the
outline of the subparts at its end, and then grayed out all but the heading that
introduced the next subpart, working down the entries until the last one,
shown here.
Principle of Capacity
Limitations:
People have a limited
capacity to retain and to
process information, and will
not understand a message if
too much information must
be retained or processed.
Figure 2.20. I again define the principle in writing and read this aloud.
Figure 2.21. I now define a key aspect of the principle, and again read it aloud.
43
44
North
Northeast
Southeast
South
Figure 2.23. This slide illustrates the first step of a demonstration to illustrate
key characteristics of our limited capacity short-term memories. I told the
audience that I would read the directions one at a time, and they should visualize a one-inch line segment pointing in each direction, with each successive
segment attached to the tail end of the previous one to form a pathway.
N
W
Figure 2.24. This slide illustrates the path that the audience should have visualized after receiving the directions in Figure 2.27. I used this and the previous
slide to teach the audience the task and to give them a little practice before
turning to the actual demonstration in the next two slides.
Northeast
Southeast
North
East
South
East
North
East
South
Northeast
Southeast
Figure 2.25. The actual test directions I used in this demonstration. The audience members were asked to close their eyes and visualize a line segment
pointed in each direction, with each line in succession connected to the
previous one to form a single pathway.
45
46
N
W
Figure 2.26. The pathway that the audience members should have visualized
on hearing the directions in Figure 2.25. Only people who noticed the
repeating patterns got this right, which illustrates that we can hold about four
units in short-term memory, and that each of those units in turn can be
composed of about four elements.
10.
Three Goals
Connect and Engage
Direct Attention
Promote Understanding
and Memory
Figure 2.27. At the end of the presentation, I first return to the overall outline
and briefly summarize the bottom line reached in each part.
47
48
1.
2.
Psychological Principles:
The Eight-Fold Way
Relevance
Appropriate Knowledge
Salience
Discriminability
Perceptual Organization
Compatibility
Informative Changes
Capacity Limitations
Figure 2.28. In this presentation, I introduce the eight principles, and as part
of the summary I briefly revisit each of them.
3.
49
50
Figure 2.29. To help the audience see in a new light, I return to the graphic
that I used at the outset. I now explain why it would not serve to communicate
well (assuming that it was ever intended to have that role, which is doubtful
in part because of the problems it would have if pressed into service in this
way).
Figure 2.30. To end on a light note, and also to underscore the generality of
the principles, I show this picture of a pizza pie, which serves to illustrate how
some of the principles could turn this into a pie graph.
51
52
Figure 2.31. I end by returning to the opening slide, which not only reminds
the audience of the topic and my identity, but also may help them stay on topic
during the question session.
1.
2.
3.
Stand to the side of the screen when you read from it, turning to
look out at the audience after reading each point.
Make sure that you arent standing in the dark. Dim the lights to the
point that your slides are easy to see, but leave enough light for the
audience to see your face and gestures (which are another way we
communicate, indicating whats most important and most relevant).
If you expand verbally on material you have presented, move
around the stage as you talk, which will keep the audiences attention on you. But stop moving as soon as you present a new slide,
video, or audio file. Dont make the audience choose between
paying attention to you versus your new material! And never pace,
which is always distracting.
53
54
4.
5.
Respond to questions.
As noted earlier, you should have prepared for questions. When you
receive them:
Dont feel that you must answer every question on the spot. Sometimes (rarely) a member of the audience hits on a point that you
(the expert) might not have considered. In such cases, if you dont
immediately know how to answer, simply say Thats an excellent
question. Rather than throwing out my first take and giving you a
glib response, let me have a few minutes to think about this.
Everyone likes being acknowledged publicly as smart, and this
response will come off much better than a rambling, nondefinitive
responsewhich could undermine your credibility in general. In
fact, you may even wish to speak to the individual after your presentation and say that youll need to do some research and will send
him or her an e-mail in a day or two to answer. Theres no shame
in not knowing everything there is to know.
Always repeat each question as soon as it is asked, so that
everyone in the entire room can hear it. Many members of the
audience may not hear questions directed at the speaker, particularly if the questioner is near the front of the room. Repeating the
questions not only ensures that everyone knows why youll
address a specific point, but it gives you a moment to gather your
wits and think about what you should say!
Close the question period by thanking the audience for their time
and attention and for giving you the chance to talk to them. And then
say Thanks again or make some other positive comment, with a
smile and a nodallowing you to leave the stage on a positive note.
55
56
Start with a bang; your very first slide should define the topic and
set the stage for your presentation.
Decide what message you want the viewers to take away from the
presentation as a wholeand do the same thing for each part and
subpart.
Provide information only when it is needed.
Dont lose your basic message by providing either too much or too
little information; be sure not to swamp the audience with irrelevant detail.
Include a slide that explicitly states every point you need to make.
In general, the amount of text you present should just be enough
to remind you to present the key points and give the audience a
way to conceptualize what youre saying.
Prepare for questions; hold some slides in reserve, in anticipation
of specific questions.
Dont ignore the audience; monitor them so that you can adjust
your presentation.
Know what you can skip.
Respond to questions, but dont feel that you must answer every
question on the spot.
Wait 20 or 30 seconds after opening the question session, and if
nobody has spoken up, make a self-deprecating comment or ask a
rhetorical question (such as, Was I skating too fast over thin ice?).
Principle of Salience: Attention is drawn to large perceptible differences. (Big relative differences grab attention.)
57
58
Dont show your audience so much text that you will have to make
the font very small.
Face the audience.
Stand to the side of the screen when you read from it, looking out
at the audience after you read each point; they can see the screen
better and hear you better.
Make sure that you arent standing in the dark.
Always repeat each question as soon as it is asked so the entire
room can hear it.
Principle of Perceptual Organization: People automatically group elements into units, which they then attend to and remember. (These
groups are easier to see and remember than the isolated components would be.)
Prepare your slides to function as your notes; the slides that you use
to jog your memory will also help the audience organize in their minds
what you say, so that they can easily comprehend and remember.
Show and tell: Use graphics in combination with text whenever
possible.
Tell the audience what you want them to conclude; as part of the
introduction, explain where you are going in your presentation and
what conclusions you are going to support.
Provide a road map: Organize your outline to reduce the load on
your audience members attention, memory, and reasoning as
much as possible.
Your conceptual structure and outline should be organized hierarchically into groups of no more than four elements.
Tell a story that helps the audience to organize the parts.
Provide concrete, specific evidence to support your conclusions.
Prepare an outline for each part that has two or more subparts.
Conclude each part and subpart with a summary.
Consider marking your progress with a banner or other signpost.
Build in breaks that allow the audience to come up for air; give
the audience time to digest what youve said.
Repeat the opening outline at the end, to remind the audience what
the presentation was about.
Prepare text and graphics to emphasize your final conclusions.
Set up a snappy ending.
Dont rush it: Dont talk so fast that the audience members are still
straining to understand one sentence as you are saying the next,
and dont whip through the slides so fast that the audience cant
take them in.
Read the most important material aloud because it will give you
the advantage of providing two waysvisual and verbalfor the
audience to store what you have to say.
59
Legible Text
61
62
1.
Use mixed uppercase and lowercase letters; dont use all uppercase letters, which are much more alike than a mixture of upperand lowercase letters (and also, who wants to be shouted at?).
Use standard font whenever possible; dont use italics for more
than a word or two.
Use normal weight as the default, and dont use bold throughout;
letters in bold are less discriminable than normal weight letters,
and also are too salient.
Legible Text
[Don't]
[Do]
2.
Dont underline.
Underlining cuts off the bottoms of letters that have descending lines
(such as p and g), which makes them harder to identify. If you want to
emphasize a word, put it in italics (provided that the letters are still easy
to distinguish in that font), make it bold, or use a more salient color.
63
64
[Don't]
Dont underline
[Do]
3.
Legible Text
signal changes in content. So dont change the color of the font
unless you want to emphasize something or distinguish among different classes of information. (For color versions of the examples
below, see insert, Figures 3.3a and 3.3b.)
When using color to signal a change, use a color for only one term
or phrase. If you make more than one word the same color, know
[Don't]
[Do]
65
66
4.
that readers will tend to group together words in the same color
even if you are only using the color for emphasis and the words are
not in fact closely related.
Avoid using more than three different colors; too many colors can
easily produce too many perceptual units and overwhelm the
viewers short-term memory capacity.
5.
[Don't]
Legible Text
[Do]
audience work harder. Letters in such fonts take more time and
energy to read because the viewer must search for the relevant lines
and try to shut out the flourishes. Even worse, such letters may also
clump together visually, making the words difficult to distinguish.
[Don't]
67
68
[Do]
6.
7.
Legible Text
[Don't]
Ensure that the font is large enough to be seen from the back of the room
[Do]
69
70
8.
Use either a serif or sans serif font, but dont mix and
mingle them arbitrarily.
Typefaces are divided into two general classes, serif and sans serif.
Serif typefaces (such as Times and Palatino) have little feet, brackets
and hooks at the ends of the lines used in the letters; sans serif
typefaces (such as Ariel and Century Gothic) have only straightforward strokes. Use the same font consistentlyotherwise
changes will be taken to convey information, and the text may look
very strange and is harder to read.
[Don't]
Figure 3.7a. If the background is too salient, viewers will have to work to
discriminate text from the background. A display should not be a haystack,
with your message the needle. If you use a patterned background, make sure
that it does not bury the text and that it is not so salient as to distract from your
message. (See color version in insert.)
Legible Text
[Do]
If fonts are so small that they are barely legible, then sans serif is
better because in a serif font, the little feet, brackets, and hooks in
different letters become inappropriately grouped together, making
the letters hard to distinguish.
If the contrast or luminance (which our perceptual systems register
as brightness) is so low that viewers can barely see the type, then
serif fonts are better because the aforementioned little feet,
brackets, and hooks provide additional cues regarding the identity
of each letter.
9.
71
72
[Don't]
if the
contrast or luminance is low, so
that fonts that should be visible
can barely be seen, use serif
fonts.
If fonts are very small, so that they are barely legible, use sans serif font;
[Do]
computers are loaded with the same font set, and some projectors
have low resolution, which can also lead to illegible type. To be safe,
use serif fonts such as Times New Roman (the most common), Garamond, or Century Schoolbook, or sans serif fonts such as Arial, Verdana, or Tahoma. I used to use only Palatino on the Mac, but I
discovered that some computers dont have this font and will substi-
Legible Text
tute Times New Roman, which changes the presentations spacing.
Printers will also substitute for fonts they dont recognize, so dont
count on your printouts looking the same if you use an uncommon
font. To be on the safe side, use True Type fonts, and an XGA (1024
768 resolution) projector. To quote President Bush again, One word
sums up properly the responsibility of any governor, and that one
word is `to be prepared.
[Don't]
[Do]
73
74
10.
[Don't]
Legible Text
[Do]
1.
2.
75
76
3.
4.
[Don't]
I.
Figure 3.11a. An opening outline should provide a general road map for the
presentation. The material above is not only overwhelming, but much of it
wont mean anything to the audience until theyve heard later parts of the
presentation.
Legible Text
[Do]
Figure 3.11b.
5.
6.
77
78
[Don't]
[Do]
Figure 3.12. As a general rule, two lines is about right for each bulleted entry.
Try reading aloud the one on the top; if its a mouthful for you, its an earful for
the audience.
Legible Text
7.
[Don't]
Socks
Twix
Snickers
Sunglasses
Mars bars
Caps
Carrots
Peas
Squash
Insect repellant
Shoe laces
Shampoo
Bread
Hand lotion
Hotdog buns
Cake
Chap Stick
Donuts
Milk
Ice cream
Figure 3.13a. If you must present every single item, organize them into categories, each of which should contain no more than four entries. But before
doing this, think again: Does the audience really need to see every one of those
individual items? Would selected examples help you to make your point more
clearly?
79
80
[Do]
Food Items
Candy
Twix
Snickers
Mars bars
Vegetables
Carrots
Peas
Squash
Baked goods
Bread
Hotdog buns
Cake
Donuts
Sundries
Wearables
Sunglasses
Socks
Shoe laces
Caps
Consumables
Hand lotion
Chap Stick
Insect repellant
Shampoo
Diary
Milk
Ice cream
Figure 3.13b.
8.
Dont change the shape of the bullets in your presentation unless you
want to signal a change in category. Viewers will assume that any
visual changereplacing round bullets with arrows, for exampleis
intended to highlight something new about the material you are
presenting.
Legible Text
[Don't]
for the
for the
for the
for the
[Do]
81
82
1.
Legible Text
Example: You could describe a new idea as innovative (which
has a positive connotation) or new-fangled (which has a negative connotation). If you are promoting the idea, the connotation
of the second term would get in your way.
2.
3.
Viewers will group labels that are placed near a graphic with that
graphic; so place labels only near graphics to which they apply.
Viewers will see labels that seem to flow from the end of a line
as part of that line, and hence assume that the labels apply to
that line. Avoid this arrangement unless the label does in fact
apply.
Viewers will group labels that are the same color as an element
with that element (e.g., green labels with green wedges, blue labels
with blue wedges, and so forth). Such grouping is confusing when
the labels dont in fact apply to the elements with the same color,
so avoid doing this.
83
84
4.
5.
6.
make the labels for the most important parts of the display more
salient (for example, by having them be relatively large, bold, a
striking color, or moving in from the side).
make the labels for increasingly smaller details, or more finegrained components, decreasingly less salient.
Legible Text
7.
eliminates the need for a key, which taxes the processing capacities of the audience members because it requires them to search
for the corresponding parts of the display;
specifies each of the relevant dimensions separately, which helps
the audience to make specific comparisons;
removes redundant labels, which clutter a display, thereby making
it more difficult to pick out the other elements.
[Don't]
Figure 3.15a.
85
86
[Do]
Figure 3.15b. Not only is the display less cluttered when direct labels are used,
but also the bars are easier to compare along specific dimensions (dwelling
type, county, year) when the labels specify the dimensions separately.
8.
Use the same size and font for labels that identify the
same types of components.
When we see a difference, we expect it to mean something and are
confused if it doesnt. Labels for each of the same type of component (wedges, bars, etc.) should use the same font and be the same
size. Entities of equal importance should be labeled with the
same font and be the same size.
Example: If you have two pictures of former CEOs and the sizes
of their names or their photos are different, the audience will
draw the obvious conclusion about your level of regard for each.
9.
Legible Text
Example: Dont label birds birds on a slide while referring to
fowl in text or in your spoken remarks.
1.
2.
87
88
3.
4.
1.
2.
Legible Text
enues over time in one part of the world with no such increases in
another), use a line graph. Using a table would force viewers to compute and compare differences among differences of numbers, which
requires work. As always, dont make viewers struggle to grasp your
message.
3.
If you have so many entries that they must be very small in order to
fit on a slide, this is a painfully clear signal that your table is not
going to be very useful for other purposes. You will probably be
better off splitting the table into two or more slides. If this wont
work, consider using a bar graph to present averages (with some
measure of variation, such as the standard error of the mean, around
each average value) rather than a table.
4.
89
90
5.
Measure 1
Measure 2
Measure 3
Measure 4
6.
14
10
12
10
Canada
15
18
14
12
Women
USA
Canada
12
9
14
10
10
10
13
12
USA
Measure 1
Measure 2
Measure 3
Measure 4
Canada
Men
Women
14
10
12
10
12
9
14
10
Men Women
15
18
14
12
10
10
13
12
Legible Text
7.
8.
1.
91
92
That said, if you need a key, consider whether you want to show the
graphic or table in your presentation; the display may simply be too
complicated to be grasped easily in this context. If you do need it:
[Don't]
[Do]
2.
Legible Text
variant of the same graphic (such as graphs or diagrams), it is centered at the top, directly beneath the title. Viewers familiar with
graphics will expect to find the key in one of these locations. If aesthetic or other considerations lead you to put the key in an unconventional location (such as in the example you just saw), realize that
the viewers may have to search for it, which will tax their limited processing capacity. However, if you consistently place the key in the
same location in a number of displays, the viewers will soon learn
where to find it.
3.
4.
5.
93
94
[Don't]
[Do]
Figure 3.17. Viewers will have to search for corresponding bars if the order of
the key does not match the order of the content elements.
Present a quick overview of a list, then gray out all but the first entry.
Ensure that only the relevant words are easily distinguishable from
the background.
Present items individually, highlighting only one at a time as you
go from top to bottom.
Use bullets only for topic sentences or specific cases.
Title a slide to focus attention on the most important information.
Use a table of numbers only when specific values are important.
Present only the information needed to make your point.
Legible Text
Principle of Appropriate Knowledge: Communication requires prior
knowledge of pertinent concepts, jargon, and symbols. (We understand and remember a message more easily if it connects to what we
already know.)
Principle of Salience: Attention is drawn to large perceptible differences. (Big relative differences grab attention.)
Principle of Perceptual Organization: People automatically group elements into units, which they then attend to and remember. (These
groups are easier to see and remember than the isolated components would be.)
95
96
If entries are labeled with pictures or icons, ensure that they evoke
the appropriate concept.
Discuss only material that is closely related to what is currently
being shown on a slide.
Dont use a table if trends or interactions are important (usually,
use a line graph).
Principle of Informative Changes: People expect changes in properties to carry information. (And we expect every necessary piece of
information to be indicated by a change in a perceptible property.)
Legible Text
If you master the principles, you wont need to memorize every one
of the recommendations any more than you need to memorize every
sentence you might say today. Just as you can generate sentences
on the basis of your (unconscious) knowledge of grammar and
words, with practice you can learn to generate good PowerPoint presentations on the basis of your knowledge of human mental
processes.
97
99
100
1.
101
102
2.
[Don't]
[Do]
[R]
[R]
[B]
[Y]
[Y]
[B]
Figure 4.1. The eye cannot properly focus on red and blue at the same time, so
a boundary should not be defined by juxtaposing red and blue regions. R is for
red, B for blue, and Y for yellow. (See this figure in the color insert.)
3.
4.
103
104
If the room will be well lit, you are better off using black figures on
a white background.
If the room will be almost dark, a white background will appear
very bright, perhaps even irritating (even if a light is shining on
you, which I recommend). So, in this situation you generally are
better off using white figures on a black background.
However, black and white are not the most exciting colors, and
you wont lose much by using different hues for figure and background as long as they are clearly discriminable. To ensure that they
are easily discriminable, be sure that the colors differ obviously in
terms of all three qualities: hue, lightness, and saturation.
5.
6.
Use one color for titles and another, less salient one
for text.
To help the viewer organize your text, use one color (e.g., yelloworange on a dark gray background) for titles and another color (e.g.,
white) for text entries beneath the title. But, as noted in Chapter 3,
ensure that the title is a more salient color than the text, thereby
appropriately directing the viewers attention.
7.
[Don't]
[Do]
Figure 4.3. The salience of colors should reflect the importance of the
headings. (See the insert for the color versions of these figures.)
105
106
[Don't]
[Do]
Figure 4.4. A red line that appears to be struggling to move to the foreground
produces an effect that is neither esthetically nor functionally desirable (see
this figure in the color insert).
8.
9.
[Don't]
[Do]
Figure 4.5. If used properly, color can be a very effective grouping device. In
this example, color in the right panel groups the appropriate scale with the
corresponding line (see this figure in the color insert).
107
108
10.
11.
For such graphics, use deeper saturations (more color) and greater
lightnesses (more light) for hues that indicate greater amounts. We
see increases in both of these visual dimensions as increases in
amount, and so these increases can signal increasing quantities
effectively.
If you must use hue to convey variations in quantity (e.g., to show
variations in temperature across a complex surface), use psychologically equal changes in color to convey equal increments in the
amountand youll need to experiment to get this approximately
right. If the psychological increments dont seem about equal,
viewers will assume that more similar quantities are conveyed by
more similar colors.
[Don't]
[Do]
Figure 4.6. Because differences in hue are not immediately perceived as differences in amount, the viewer is required to memorize a key if hue alone is used
to represent quantities (as in the version on the left). However, if the saturation and lightness vary along with hue, as in the version on the right, the
viewer can easily see differences in represented amounts (see this figure in the
color insert).
12.
109
110
[Don't]
[Do]
Figure 4.7. The display on the left uses size to indicate consumption by region,
hue to indicate income, and saturation to indicate temperature; it is a puzzle
to be solved. The display on the right uses only size and saturation to convey
amounts, and viewers can clearly sense the orderings here (see this figure in
the color insert).
Blue
Red
34
28
1
20
43
0
5
42
0
6
9
76
3
3
41
16
8
1
1.
111
112
[Don't]
[Do]
[Don't]
[Do]
Figure 4.9. Line spacings that differ by at least a 2:1 ratio are immediately
discriminable; if regions are not immediately discriminable, the viewer has to
work harder than necessary to compare corresponding elements.
Figure 4.10. An annoying shimmer occurs when your visual system is struggling to detect a poorly defined edge.
2.
113
114
1.
[Don't]
[Do]
Figure 4.11. If salience leads readers to see the house before the data, they will
have to work to sort out the content.
2.
3.
[Don't]
[Do]
Figure 4.12. If background elements group with content, the display is likely
to be misinterpreted.
[Don't]
[Do]
Figure 4.13. The editorial content of a background can allow you to underline your message effectively, or it can create confusion; the background
pattern should not conflict with the message of the display.
115
116
1.
2.
Even if you are not discussing a sequence of events, you can use
motion to guide attention: In general, presenting a label first and
then having an arrow extend from the label until it touches the
labeled part is an effective way to direct the viewers attention.
3.
4.
117
118
5.
6.
7.
8.
9.
10.
1.
119
120
2.
3.
4.
Use sounds that are appropriate for the topic and point
being made.
If you use a sound to introduce a slide, ensure that it is compatible
with the point.
Example: If you want to introduce a presentation or section on
the effects of a hurricane, the sounds of a ferocious wind might
be appropriate, especially if accompanied by a video clip of the
wind ripping through a town. The sounds of gentle rain on the
roof would not be appropriate.
5.
6.
However, you should use sound in this way only very sparingly,
and your timing must be just right. Otherwise you will not seem
serious, and the audience will discount your message.
7.
121
122
8.
If the sounds are discrete, such as the sounds different animals make,
show a photo of each object, and click on the graphic to present the
corresponding sound (assuming that this is relevant to your message).
9.
10.
In addition, dont make the sounds too loud, which will not only cause
the audience members to cringe, but may lead them to anticipate
further assaults and thus not pay close attention to your message.
Principle of Salience: Attention is drawn to large perceptible differences. (Big relative differences grab attention.)
Principle of Perceptual Organization: People automatically group elements into units, which they then attend to and remember. (These
groups are easier to see and remember than the isolated components would be.)
123
124
Principle of Informative Changes: People expect changes in properties to carry information. (And we expect every necessary piece of
information to be indicated by a change in a perceptible property.)
Communicating Quantitative
Information: Graphs
125
126
1.
2.
3.
1.
127
128
[Don't]
[Do]
Figure 5.1. Do not use a scale with pie graphs; the viewer will have to struggle
to count the number of ticks. In this graph, credit cards and revolving credit are
the topic of interest, and hence this information is most salient.
2.
3.
[Don't]
[Do]
Figure 5.2. Use an exploded pie only when a distinct contour can be disrupted
by exploding the wedges; the two intact pieces of the pie on the right define
a contour, which makes the exploded wedge stand out.
4.
129
130
[Don't]
[Do]
Figure 5.3. If too many wedges are exploded, none stands out.
5.
6.
In general, order from smaller to larger, with size increasing clockwise. Because values on a clock face increase in a clockwise direction, we expect greater quantities to be indicated by greater arcs in
a clockwise direction.
However, if you wish to emphasize the larger components, put the
largest at the top or in the 1:00 position (the first that the viewer
will focus upon when scanning in a clockwise order), and arrange
the other wedges in decreasing size from this anchor point.
[Don't]
[Do]
(left) compared with Italy in particular (right)? (These data were cited
in 1991, prior to the formation of the European Union.) This is difficult to fathom, in part because the relevant wedges are in different
locations in the two piesa shift that cannot be avoided when the
proportions vary widely.
Now compare the two Do pie graphs; it is clear that there is a
rough correspondence between the revenues per region (left) and
employees per region (right). It is easy to compare multiple pies
when the wedges are in roughly the same positions in each. But, as
illustrated in Dont, when the corresponding parts are in different
locations, the audience is forced to search for them one at a time
which requires effort. Our brains have limited processing capacity,
and if forced to strain, many audience members may simply give up.
(Also note that because Western Europe was of greatest interest to
the designers, the largest wedge is put at the beginning of the clockwise progression.)
If the proportions are very different across levels, and the audience
is supposed to compare specific components, a bar graph is the
preferable format.
131
132
[Don't]
[Do]
Figure 5.5. Corresponding wedges are hard to compare if, as in Car Markets,
they are not in corresponding positions.
1.
2.
3.
133
4.
50
Number of Employees
134
40
30
20
10
Men
Women
Sales
Men
Women
Research
Figure 5.6. This bar graph could easily be turned into a visual table because
the icons neatly label the bars, making the labels on the X-axis superfluous if
the context in which the display is presented is clear enough. In this graph,
each icon represents five employees, but the unit size is determined both by
the ease of identifying the icons (if they are too small, they will not be discriminable) and the precision of the data.
5.
[Don't]
[Do]
Figure 5.7. The icons should reflect the subject matter, which they do in both
panelsbut the appearance of the pattern is incompatible with the message
in another way: More is seen as moreso longer bars or greater numbers
of icons should specify larger amounts of the relevant substance or more
frequent occurrences of the relevant event. In Dont,greater numbers of
icons actually indicate a lower crime rate! Do not convey less by more.
135
136
6.
7.
[Don't]
[Do]
Figure 5.8. Pictures in a visual table should provide another kind of label and
should not interfere with the interpretation of the display.
[Don't]
[Do]
Figure 5.9. It is much easier to compare extents at the same orientation than
at different orientations.
8.
137
138
[Don't]
[Do]
Figure 5.10. The line graph clearly indicates the number of new accounts, the
average deposit, and the total deposits per season, as well as the relations
among these measures. The same information is embedded in the four panels
above, but considerable effort is required to decode it. We perceive height and
width as specifying an area, not as separate variables.
1.
[Don't]
[Do]
Figure 5.11. The continuous variation of a line is compatible with the continuous variation of time; if you want the viewer to note precise point values,
put dots or symbols along the line.
2.
3.
139
140
[Don't]
[Do]
Figure 5.12. Experienced graph readers can interpret typical patterns of lines
at a glance. Hence, it is goodespecially in a presentation when viewers may
not have much time to decode a displayto make use of familiar patterns to
convey interactions.
4.
[Don't]
[Do]
Figure 5.13. The heights of bars define specific points, whereas lines specify
continuous variations. It is more difficult perceptually to break up a line into
points than to detect the tops of bars.
5.
141
142
6.
[Don't]
[Do]
Figure 5.14. If the focus is on NBC, the graph on the right is preferable; the
increased salience of the line for NBC immediately draws the viewers attention.
7.
8.
9.
[Don't]
[Do]
Figure 5.15. The points on the lines are especially important and should be
emphasized by discriminable dots.
143
144
[Don't]
[Do]
Figure 5.16. Studies have shown that the four lines and symbols on the right
are highly discriminable.
10.
11.
[Don't]
[Do]
Figure 5.17. The staggered heavier grid lines help viewers to locate specific
values on the axes for specific data points.
Figure 5.18. Without an inner grid, it is almost impossible to tell that the lines
have the same difference in value on the Y axis over the 2 and 5 values on the
X axis.
145
146
1.
2.
3.
[Don't]
[Do]
Figure 5.19. A line graph for these data inappropriately suggests a rapid rise
along a continuous variation.
4.
5.
147
148
[Don't]
[Do]
Figure 5.20. The trend toward decreasing numbers of defects with increasing
production, especially for the industrial widget, is not immediately evident
from the mixed line/bar display; the visual system more easily groups different
lines into a pattern than it groups lines and bars.
6.
7.
[Don't]
[Do]
[Don't]
[Do]
Figure 5.22. A gratuitous change of order makes it difficult to compare corresponding bars.
149
150
8.
[Don't]
[Do]
Figure 5.23. Making one element very salient will put it at the center of attention. If you want to emphasize that element, this is appropriate; but if you
are simply trying to provide visual variety, as here, it is likely to confuse the
audience.
9.
10.
[Don't]
[Do]
Figure 5.24. Six perceptual units are too many to apprehend immediately;
proper grouping not only makes the display easier to take in but also helps to
group the bars with their labels.
[Don't]
[Do]
Figure 5.25. If the Y axis is too short (Dont, at left), the viewer cannot immediately estimate the price of tea in 2010.
151
152
1.
2.
[Don't]
[Do]
Figure 5.26. A combined step graph is difficult to read and easily mistaken
as illustrating a cumulative total.
3.
4.
Fill the area under the line with a single pattern or color.
The primary virtue of step graphs is that they produce patterns that
indicate specific trends. As is evident in Do, if the area under the
line is filled with a color or texture that is different from the background, the grouping law of similarity (as summarized in the
Appendix) will lead the viewer to see that area as a single
shape, facilitating recognition of a trend. But only a single color or
texture should be used; otherwise, the impression of a single region
will be disrupted.
153
154
[Don't]
[Do]
Figure 5.27. Making one step wider than the others gives it unequal
importance.
[Don't]
[Do]
Figure 5.28. The shape of a series of rising and falling steps is meaningful
and should not be disrupted.
1.
2.
data from two categories are intimately related (and so the fact that
the points cannot be distinguished is itself compatible with the
message being conveyed);
data from two categories are clearly distinct (as evident when the
clouds formed by different sets of points can be easily distinguished because they are in different parts of the display).
155
156
[Don't]
[Do]
[Don't]
[Do]
[Don't]
[Do]
3.
Multiple Panels
When should you divide the data up and display it in separate
panels?
157
158
1.
[Don't]
[Do]
Figure 5.32. A display containing more than four perceptual units cannot be
apprehended in a glance. Break such displays into separate panels.
2.
3.
[Don't]
[Do]
[Don't]
[Do]
Figure 5.34. Plotting two related variables together can reveal useful information, such as the impact of equipment failure on the amount of oil that was
pumped.
159
160
Use concepts and display formats that are familiar to the audience.
Principle of Salience: Attention is drawn to large perceptible differences. (Big relative differences grab attention.)
Principle of Perceptual Organization: People automatically group elements into units, which they then attend to and remember. (These
groups are easier to see and remember than the isolated components would be.)
161
162
Principle of Informative Changes: People expect changes in properties to carry information. (And we expect every necessary piece of
information to be indicated by a change in a perceptible property.)
Communicating Qualitative
Information: Charts, Diagrams,
Maps, Photographs, and Clipart
163
164
I begin with general recommendations, then turn to specific recommendations for each type of display.
1.
2.
[Don't]
[Do]
Figure 6.1. A diagram intended to show the drivers position relative to key
controls is easier to read if the irrelevant components are eliminated and the
relevant ones are highlighted.
3.
165
166
If you even suspect that your audience may be in doubt about the
meanings of the symbols you are using, state those meanings
explicitly.
Example: When describing a flowchart, indicate whether different
shapessuch as squares and diamondshave different meanings
(such as an operation versus a decision/branching point).
4.
5.
6.
Similar shapes
Similar colors
Similar shadings
Similar sizes
Similar line weights
A line or arrow that links objects
Nearby locations, which are clearly distinct from other clusters
[Don't]
[Do]
Figure 6.2. A complex chart can be sorted out relatively easily if it is visually
organized into components.
7.
167
168
8.
9.
Charts: 6 Pointers
Charts can make complex sets of information clear at a glance, or
can make complex information seem even more complex. For
instance, consider one chart that was created in the wake of Hurricane Katrina. This hurricane was no laughing matter, nor was the
feeble response of the Federal Emergency Management Agency
(FEMA). Nevertheless, Jon Stewart, the host of the Daily Show
(broadcast on Comedy Central), couldnt help commenting on a chart
released by FEMA (during the week of 26 September 2005), Figure
6.3. He said: What should FEMA have done? Perhaps the answer
can be found on their website. . . . This chart [shown below], clearly
depicting the agencys responsibilities in the event of a disaster. . . .
It begins with a response to a disaster, leads to recovery, mitigation,
risk reduction, prevention, preparedness . . . (dramatic pause) and
Figure 6.3. The actual chart from the FEMA website. Can you understand what
it means?
169
170
1.
2.
3.
Company Divisions
Consulting
Publishing
People can read such a list faster than they can decipher a chart, and
a short list will tax short-term memory less than would a multielement display. However, if you want the viewers to compare
numerous different combinations of entities (such as the relations
4.
5.
[Don't]
171
172
[Do]
Figure 6.4b.
[Don't]
[Do]
Figure 6.5. Every important piece of information, whether entity or relationship, should be identified in the display.
6.
All else being equal, respect the convention in Western culture and
make a sequence over time proceed from left to right.
However, respect the specific conventions of your audience. For
example, computer programmers have developed a specialized
subculture, where flowcharts are often presented from the top down.
Identify relationships.
Identify every important piece of information in a display, as in the
Do version of this family tree. (Figure 6.5).
Diagrams: 8 Recommendations
Diagrams are pictures of objects or events that use both pictorial elements and symbols (such as arrows to show movement) to convey
informationto show the wiring in your kitchen, the way the Eiffel
Tower sways in a strong wind, or the assembly of your new barbeque. Unlike charts, parts of diagrams must resemble the things
they represent.
1.
2.
3.
173
174
[Don't]
[Do]
Figure 6.6. If spatial relations are distorted to show components, the distortion should be easy to reintegrate; otherwise, readers will have to expend
effort to see the components in place.
4.
5.
6.
Ensure that the order and way in which parts are added, removed,
or modified (e.g., shown to bend) mimics the actual order or way
in which they would be added, removed, or modified. This technique allows you to discuss each relevant part before turning to
the next, and the use of animation will draw the viewers attention
to the next topic of discussion.
7.
8.
Similarly, never have one part of the display drawn in two dimensions with a three-dimensional part tacked on to it.
175
176
1.
2.
3.
4.
177
178
[Don't]
[Do]
Figure 6.7. A map that includes too much extraneous detail for its purpose is
not helpful; to specify the location of the firehouse, only the efficient routes are
necessary.
5.
[Don't]
[Do]
6.
7.
179
180
[Don't]
[Do]
Figure 6.9. If both the width and height of the triangles are varied, we see
neither variable very well; it is far better to use two different displays, one for
each variable.
8.
9.
[Don't]
[Do]
[Don't]
[Do]
181
182
10.
If the grid lines are spaced every half inch on the map, the scale
should indicate the distance in terms of this unit length or in terms
of a familiar multiple of this increment, say, 1 inch. Its often most
useful if the scale indicates a unit that corresponds to the distance
between heavy grid lines as in Do in Figure 6.12.
[Don't]
[Do]
Figure 6.12. If the scale does not line up with the increments of the grids,
mental division is necessary to use it; spare the viewer this effort by aligning
the scale and grid increments.
11.
[Don't]
[Do]
Figure 6.13. Making the major roads correspond to more salient lines will
help viewers to organize the map as well as to find the most efficient routes.
12.
1.
183
184
2.
3.
4.
5.
6.
7.
8.
185
186
9.
10.
Use symbols, concepts, and jargon that are familiar to the audience members.
Principle of Salience: Attention is drawn to large perceptible differences. (Big relative differences grab attention.)
Ensure that all aspects of the graphic are clearly discriminable from
the background.
Ensure that shapes of meaningful regions are easily identifiable.
Ensure that photos and clipart do not become too grainy when
inserted into the slide.
Principle of Perceptual Organization: People automatically group elements into units, which they then attend to and remember. (These
groups are easier to see and remember than the isolated components would be.)
187
188
189
191
192
193
194
195
196
Content Templates
Content templates not only provide formats, color schemes, and
graphics but also actually provide suggested text for specific subjects (I quote from PowerPoint Help). For example, there are templates that supposedly help you in a brainstorming session, to
communicate bad news, and so on. Want to seem insincere?
Thoughtless? You run the risk of getting such reactions by using
these canned thoughts and ideas.
In addition to including uninspired and uninspiring pointers, some
of these templates include bad formatting. For example, in
Employee Orientation bullet points fly in one-by-one, which can
draw attention but also prove a distraction. And in Selling a
Product (for the Mac), the background pulses, which is interesting
but very distracting (this feature was fixed in the Presentation on
Product or Service template in Office 2007). Furthermore, many
templates have default color schemes that reuse the same colors (or
very similar colors) to mark different content elements (thereby
producing a violation of the Principle of Discriminability).
Other templates (such as Selling a Product) contain exotic conceptual diagrams, and the user is expected to fill them in (violating
the Principles of Appropriate Knowledge, Compatibility, and
oftenInformative Changes).
Pie Charts
The PowerPoint program prevents you from modifying pie charts
(i.e., pie graphs) in several ways that would improve them. For
example, in the Mac version you cannot explode one, and only one,
wedge. Its either all or nothing, which violates both the Principles of
Informative Changes and Discriminability. It also does not place
wedge labels inside the pie, which violates the Principle of Perceptual
Organization. (Both of these problems have thankfully been fixed in
Office 2007 for the PC.)
In addition, the program has options and default settings that get
in the way of making effective pie graphs. For example, when making
black-and-white pie graphs, dot spacing is used to distinguish the
parts rather than hatching or different gray-scale values; this practice violates the Principle of Discriminability. Office 2007 does use
different gray scales, but they are not necessarily discriminable.
The PowerPoint program also allows the unwary user to create a
donut where the center of a pie is removedthis violates the Principle of Appropriate Knowledge and could confuse more than illuminate. The Program also can create a pie of pie and a bar of pie
in which some of the values are extracted from the original pie and
put into a new pie or bar connected by lines to the old one. Because
the whole is split across two displays, this can easily be misleading
and confusing. What audience could immediately grasp such
displays?
Line Charts
Next, consider Line Charts (i.e., line graphs) and scatterplots. The
default in the PowerPoint program does not label lines directly (but
instead uses a key), which forces you to do so by hand or to use the
key (which would thereby force the viewer to memorize the key and
then search for the corresponding lines). The program also does not
by default indicate a discontinuity in a scale via slash marks or a
zigzag, which may thereby violate the Principle of Compatibility
(because more may not look like more). And it also doesnt indicate
when points overlap in a scatterplot; one is simply hidden by the
other, which is a violation of both the Principles of Informative
Changes and Compatibility. In addition, many of the default settings
in the PowerPoint program are not optimal. For example, axis category labels are made bold but the axis value labels are not, which
violates the Principle of Informative Changes, and the weight of the
line illustrating the content line is 2 point, whereas the weight of the
197
198
Bar Charts
Bar Charts (i.e., bar graphs) not only have the same problems with
axes and labeling that exist for line graphs but have additional ones
as well. For example, they can display one variable as bars and the
other variable as area, such that the amount of shaded area is relevant to one variable but not the other. This violates not only the Principle of Appropriate Knowledge but also the Principle of Capacity
Limitations (forcing people to work hard to compare the two types of
variables). In addition, it can substitute cones for bars (included
cones with colors to indicate different layers within them), which
makes it difficult to tell what values are being depicted.
The PowerPoint program also can show more than one row of
three-dimensional bars on the same graph such that one row hides
the otherthereby not signaling information with visible changes.
You can avoid this option, but the fact that presenters can easily
stumble into using it is unfortunate.
Three-Dimensional Displays
The addition of the third dimension to a graph is a mixed blessing.
For many purposes, it does no harm and makes the presentation
more interesting (see the literature review in my book, Graph Design
for the Eye and Mind). However, for other purposes it disrupts communication by obscuring the precise amounts displayed (such as
occurs when the tops of bars are not clearly aligned with values on
a wall that serves as the Y axis). With such concerns in mind, it is
worth approaching with caution the following characteristics of the
programs three-dimensional displays:
They can change the elevation of ones vantage point from the 15degree default to less than 90 degrees (a view from below) to more
than 90 degrees (a view from above). A view from above will not
show relative heights, and thereby obscures differences in
amounts being shown.
The program allows you to rotate the graph in any direction, which
can completely obscure differences between bars and lines.
Because of the angle at which the axes meet, the ratio of the X axis
to the Y axis can produce misleading patterns.
199
200
Appendix: Psychological
Principles and Their
Specific Aspects
201
202
Relevance
The Principle of Relevance: Communication is most effective when
neither too much nor too little information is presented.
People unconsciously try to organize information into concepts
or narratives. Its almost as if we have an urge to understand or an
urge to explain. If you dont provide enough information to make
your case, the audience will be left with fragments of the story and
will flail about, trying to fill in the missing pieces (assuming that
they are motivated; if not, they will probably give up and tune you
out). Less obvious, because we try to fit related material into a single
concept or narrative, you can also confuse by presenting too much
information. Again, the audience will try to fit all of the pieces
together, and additional details will require additional effort. The
audience needs to know whats the foreground and whats the backgroundwhats important and whats context. If you present too
much, the audience may have trouble making this distinction. So,
before beginning to prepare your presentation, you need to decide
what message you want to convey; only after you have made this
decision can you determine what information to include.
Appropriate Knowledge
The Principle of Appropriate Knowledge: Communication requires
prior knowledge of pertinent concepts, jargon, and symbols.
After you have decided what to say, you need to determine how to
say it. You will communicate effectively only if you make use of what
the audience members already know. Ideally, rely only on concepts,
jargon, and symbols that are familiar to your audience; if novel concepts, jargon, or symbols must be used, define them. But more than
that, respect the fact that we humans understand new information by
relating it to information previously stored in memory. To understand
these words, for instance, you need to have stored their meanings
previously. As a general rule, the audience will understand a new idea
most easily if it is presented as growing out of a familiar one. For
instance, if you are presenting a plan to market air conditioners in
rural China to an audience that is very familiar with marketing products in international markets, you might draw an analogy to how a
previous market was opened (perhaps rural Australia)or how a similar product (perhaps electric heaters) broke into the same market.
Appendix
Salience
The Principle of Salience: Attention is drawn to large perceptible
differences.
We are led to focus first on the most prominent features of a display, those that are brighter, darker, larger, in motion, or in some
other way clearly different from their surroundings. This ability is
very important, both for our current survival and that of our ancestors. Thus, its no surprise that an ancient part of the brain, the superior colliculus (for vision), produces a kind of attentional reflex. This
part of the brain is functional almost immediately after birth, whereas
the parts of the brain required to shift attention voluntarily become
functional only a month or two after birth. As a consequence of the
relative ages when the two sorts of mechanisms come online,
young babies are subject to attentional capture: A flashing light
might grab their attention, and then they are stuck looking at that
glittering object until something else grabs their attention. The same
is true for sound: We reflexively pay attention to louder or differentpitched sounds.
The key to salience is not the absolute properties of a sight or
sound, but its properties in comparison to those of the background
or other elements. When you read THIS the word this is salient;
BUT WHEN YOU READ THIS THE WORD THIS IS NOT.
Discriminability
The Principle of Discriminability: Two properties must differ by a
large enough proportion or they will not be distinguished.
This principle derives from basic characteristics of how neurons
(i.e., brain cells) interact. Consider neurons that detect spots of light
in the first portions of the cerebral cortex to receive input from the
eyes. Nearby neurons tend to inhibit each other: If one is strongly
activated, it tries to turn off its neighbors. This arrangement
apparently plays a role in specifying the locations of edges, say,
between a piece of white paper and the unpainted wooden table
top on which it rests. The inhibition allows the system to sharpen up
the boundary between the two regions (paper and table top),
because the neurons that are stimulated more by the light bouncing
off the paper will inhibit those on the other side of the edge. However, if the paper were on a white table top, it would be much more
difficult to see it, its edges would not be so well defined. Similar
203
204
Perceptual Organization
The Principle of Perceptual Organization: People automatically group
elements into units, which they then attend to and remember.
The world is not like the diagram of a cow on a butchers wall:
There are no dotted lines that specify how to organize it. Rather,
our brains must produce this organization. So important is this task
that numerous different mechanisms have evolved to accomplish it.
In the text, I focused on just the grouping laws but sometimes had
to refer to other aspects of the principles that underlie the way we
organize visual input. I here summarize all of the relevant
principles.
Grouping Laws
The visual system automatically groups perceptible elements into
perceptual units. Because a unit is a psychologicalnot a physical
entity, we cannot determine whether a slide has too much information simply by counting marks on the screen. Rather, we must
understand how perceptual organization works.
Proximity
People group together nearby marks. For example, you see xxx
xxx as two groups, whereas you will see the same number of elements presented as xx xx xx as three groups. Thus, for instance,
putting labels in tables closer to the rows and columns that they label
than to anything else will automatically group them with the proper
column. Similarly, pausing in the middle of a sentence separates
words in time, which leads listeners to organize them into separate
groups.
Appendix
Similarity
People group similar marks. For example, you see - - - -IIII as two
groups, and --II--II as four groups (or, as two groups with two
groups within each). This method of grouping plays an especially
important role when different colors are used in a display; by
matching the colors of items and their labels, the two will automatically be grouped together.
Good continuation
People see segments that line up on a smooth curve as a single unit.
For example, you see - - - - - - - - as one unit, not eight separate
dashes. In contrast, you see - - - - - - - - as two units, because the
dashes now line up as two smooth functions.
Common fate
People group together marks that are moving similarly. For instance,
if you have material glide in from one side of a slide, viewers will
group parts of the same word or figure that move at the same rate
and at the same time.
Good form
People group marks that form simple shapes, such as [ _ ], which we
see as one shape; in contrast when the same marks are arranged so
that they do not form a simple shape, ] [ _ , we see them individually.
Input channels
What we see comes to us from a number of input channels, which
you can think of as analogous to different lenses on a camera (they
focus on smaller or larger regions, with higher or lower resolution,
respectively). These channels differ partly in regard to the size of the
region they register, which also affects the amount of detail they pick
up. Less detail is detected when larger regions are monitored. For
instance, when seen from 10 feet away, one channel might register
objects as large as a car, but not much detail; another might register objects as large as a motorcycle, but with more detail; another,
objects as large as a TV set, with even more detail; and finally,
another might register objects as large as a cell phone, in great
detail. Which channel is best? That depends on the task at hand.
For walking around a car, output from a channel that picks up large
shapes would be most helpful; for threading a needle, output from
a channel that registers fine detail would be most helpful.
The acuity (level of detail) of a channel is described by its spatial
frequency, which is the number of regular light/dark changes (for
example, light and dark stripes of equal width) that fit into a specific
amount of your visual field. Each channel responds to a range of
205
206
Spatial imprecision
Different neural pathways register shapes and spatial relations, and
the outputs from these pathways are not always combined properly
or precisely. We register precise spatial relations (e.g., between parts
of an exploded diagram) only with effort. External aids, such as grid
lines in a graph, can help the viewer apprehend spatial relations
among shapes precisely.
The various aspects of the Principle of Perceptual Organization are
summarized in Table 1.
Appendix
Table 1. Aspects of the Principle of Perceptual Organization
Aspect
Specific Cases
Explanation
Example
Grouping Laws
Proximity
Nearby marks
are grouped
together
Similar marks
are grouped
together
Marks along
a smooth function
are grouped
together
Marks moving the
same way are
grouped together
Marks that define a
simple figure are
grouped together
XXXX vs. XX XX
Similarity
Good
continuation
Common fate
Good form
Input Channels
Acuity
Spatial
Imprecision
-------- vs.
-------
..
vs. . .
( ) vs. ( __
--------------versus
- - - - - -___
-___
___
The hour hand
of a clock at 1:00
versus at 1:30
Separable
Integral
Dimensions that
are not seen
independently
The size of a
circle and the
orientation of a
radius within it
The height
and width of a
rectangle
Orientation
Integrated
versus
Separated
Dimensions
We do not always
precisely organize
objects in terms of
their spatial
relations
The relation
between a line in
a graph and a
point on the
Y axis is not easy
to see precisely
Compatibility
The Principle of Compatibility: A message is easiest to understand
if its form is compatible with its meaning.
This principle also has several distinct aspects, as follows.
207
208
Appearance-Meaning Correspondence
What you see should be what you get. The Stroop phenomenon (discussed in Chapter 1, where people have difficulty naming the color
of ink used to print words that name colors when the ink color is different from the color name) is an extreme example of what goes
wrong when this principle is violated.
More Is More
Greater amounts along a visual dimension are interpreted as representing larger amounts of a measured substance or quality. For
instance, larger wedges in a pie are seen as representing greater
amounts, as are higher bars or higher points on a line.
Cultural Conventions
All cultures produce common associations between some visual
characteristics and concepts. For example, in Western culture,
increases commonly are indicated from left to right or in a direction
clockwise around a circle. But these associations are not shared
worldwide, and the conventions of your audience should not be contravened (e.g., during the height of the Cultural Revolution in China,
some firebrands wanted to reverse the meaning of red and green in
stoplights, making red mean go).
Perceptual Distortion
The eye and mind are sometimes prone to distort the actual physical
form we perceive and may distort in systematic ways. Thus, compatibility must be considered not simply in terms of the physical
characteristics of an object or event, but instead in terms of how
those characteristics are perceived. Some visual dimensions are systematically distorted, most important:
Appendix
The aspects of the Principle of Compatibility are summarized in
Table 2.
Explanation
Example
AppearanceMeaning
Correspondence
The interpretation of a
pattern should not
conflict with its meaning
A penguin should
not be used to represent
birds in general
More Is More
Cultural
Conventions
The interpretation of a
pattern should be
consistent with its
cultural interpretation
Perceptual
Distortion
We dont always
accurately register what
we see or hear
Area is underestimated;
vertical lines appear
longer than horizontal
ones
Informative Changes
The Principle of Informative Changes: People expect changes in
properties to carry information.
When things stay the same, there is no new information; when
something changes, there is or should be new information. If a
noticeable change in the appearance or the sound of your presentation doesnt mean something, the change is simply a distraction. In
addition, every piece of necessary information should be signaled
by a perceptible change visible marks or audible sounds. Suitable
links and labels should be shown, as should signs indicating that a
different sort of information (e.g., projections of future trends, or different types of measurements) is being presented.
Finally, when the same word or illustration conveys more than
one meaning it is, by definition, ambiguous. The audience should
not have to struggle to resolve ambiguity, in either words or
graphics, but rather should be provided the information as directly
and transparently as possible. You can specify the appropriate
meaning of an ambiguous word either by paraphrasing it or
providing the appropriate context (e.g., the word port is not
ambiguous in the context of boats or wines). Ambiguity in graphics
209
210
Capacity Limitations
The Principle of Capacity Limitations: People have a limited capacity
to retain and to process information, and so will not understand a
message if too much information must be retained or processed.
This principle has two parts, Memory Limitations and Processing
Limitations.
Memory Limitations
To follow a presentation, the audience must not only keep in mind
what they are seeing and hearing at each moment but also must
recall earlier material in order to integrate each new piece into the
emerging overall picture. These requirements present a challenge to
the speaker because both our ability to hold information in mind and
to store information in our memories is limited. However, you can
work surprisingly well within these limitations if you organize and
present your material properly.
Hierarchical Organization
More fundamentally, heres a crucial point about short-term memory
that can be exploited generally: Although we can only hold about
four units in mind at once, each of those units itself can include four
units. Thus, organizing material hierarchically can vastly improve the
ability to hold information in mind.
Appendix
near future, such as Japan, France, Pakistan, Norway, Italy, China,
Egypt, Mexico, Peru, and Singapore, you would tend to remember
the countries at the beginning of the list (such as Japan and France)
and those at end of the list (such as Egypt, Peru, and Singapore). We
hold the last four or so items in short-term memory, but must store
the first few items in long-term memory, the relatively permanent
store where information is retained even when we are not consciously aware of knowing it. Because short-term memory is too limited to hold all the items at once, we need to store the initial items in
long-term memory. And because we have more time to think about
(process) the items we get earlier in the sequence, they are more
likely to have been stored than the items in the middle of the
sequence. Thus, the middle items are especially vulnerable to being
forgotten; they are not in short-term memory and the person hasnt
processed them enough to store them in long-term memory.
The bottom line: Put the most important material at the beginning
or end of a section (which is one reason I recommend summarizing
at the ends of sections).
Thinking Is Learning
An extraordinarily effective way to help your audience learn your
material is to lead them to think about it. The more people think
through facts or ideas, the more likely they are later to remember
this material. For example, in one study, people were asked to visualize scenes described by sentences and simply to rate how vivid
each mental image waswith no warning that they would be asked
to recall the sentences. When these people later received a surprise
memory test, they did just as well as another group who were told
from the outset that they should try to memorize the sentences. The
more someone is led to work through an idea, the more likely he
or she is to store it in long-term memoryand hence to be able to
remember it much later.
Multiple Memories
Finally, there is more than one type of long-term memory, and retention is vastly improved if people are led to store information in more
than one type of memory. In particular, people store words and the
211
212
Processing Limitations
Humans have only a limited ability to process information.
Effortful Search
Searching a display requires effort, and if too much effort is required
the viewer will give up. So vary salience in order to guide the viewer
through a display.
Tiring Transformations
Asking the audience to add, subtract, or average values is a sure way
to lose them. People do not like to expend effort, especially if they are
not sure what the payoff will be.
The various aspects of the Principle of Capacity Limitations are
summarized in Table 3.
Specific Cases
Explanation
Memory
Short-term memory We can hold only 4 groups
Limitations limitations
of information in mind
at the same time
Chunking
Individual items can be
organized into units, each
of which can contain up
to 4 units and still be held
in short-term memory
Privileges of the
The first and last entries
First and Last
in a sequence are
remembered better than
entries in the middle
Hanging It on
Memorys Hooks
Information is learned
more easily if it is a
variation of familiar
information
Example
We can keep in mind
4 entries in a list
Hierarchical
organization
Appendix
213
Table 3. Continued
Aspect
Specific Cases
Explanation
Example
Thinking Is Learning
Searching for
corresponding labels
in multiple panels
Asking viewers to find
mean values from
entries in a table
Multiple Memories
Processing
Limitations
Effortful Search
Tiring
Transformations
Index
slides and, 34
with sound, 11920
titles, 87
with transitions, 11617
Audience
breaks for, 24
conclusions of, 3032
connecting with, 46, 18
directing attention of, 2122
facing, 53
graphs and, 127
knowledge, 2628
monitoring, 5253
photographs and Clipart and,
185
sounds and, 121
speaking to, 5253
specific, 2021
thanking, 55
time given to, 76
visualizing, 44f, 45f
216
Index
Index
Coming up for air, 24, 121
photographs and Clipart and, 185
Common fate, 205
Communication options, 2324
Compatibility, 910, 15, 18, 45, 58, 96,
119, 124, 16162, 18788, 193,
2079
aspects of, 209t
of colors, 10911
Complex displays, 11819, 192f, 193
components of, 16667
Concepts, 5
Conclusions
of audience, 3032
effective, 4950
graphical emphasis of, 4849
of subparts, 39
supporting, 3739
Connecting, with audience, 46, 18
Connotations, 8283
Content templates, 196
Continuous scale, 140
Conversational speaking, 52
Cultural conventions, 208
Dashed lines, in line graphs, 143
Data
graphing, 126
in multiple panels, 15860
Delivery, 5155
Demonstrations, preparing, 4247
Denotations, 8283
Design templates, 19596
Diagrams, 16469, 17376
animating, 175
compatible movement, 174
components of, 17374
exploding, 173
parts of, 175
static three-dimensional, 175
structures illustrated by, 173
Dictionary, 5
Dimensions, integrated v. separate, 206
Discrete points, in line graphs, 14344
Discriminability, 8, 18, 40, 57, 95, 123,
161, 187, 194, 197, 2034
217
218
Index
Graphics (Cont.)
returning to, 50f
saturation for, 1089
Graphs, 31f, 32f, 12627
absorbing information on, 39f
audience-friendly, 127
effective, 40f
line, 13746, 19798
pie, 12732, 197
for relative amounts, 126
for specific data, 126
step, 15254
tables v., 31
Graying it out, 7, 74, 75
Green, red and, 1023
Grid lines, 9091
Grid markings, in maps, 180
Grouping laws, 2045
Hallock, Joe, 110
Height
of location markers in maps, 179, 180f
in visual tables, 137
Hemingway, Ernest, 5
Hierarchical organization, 7980, 210
labeling, 8586
Hue, 100
measurement specifications and,
1089
Icons, in visual tables, 13435
Informative Changes, 1011, 15, 18, 58,
96, 124, 162, 166, 188, 193, 197, 198,
20910
of design templates, 195
Inner grid, in line graphs, 14446, 145f
Input channels, 2056
Integrated dimensions, 206
Interactions
line graph portrayal of, 13940
tables and, 89
Introduction, 2533
Italics, 62
Jargon, 16566
Jokes, 1011, 25, 50, 82
Index
information in, 138f
inner grid in, 14446, 145f
interactions displayed by, 13940
interpreting, 140f
point values and, 141
salience in, 142
X axis of, 139, 141
Y axis of, 141
Logarithms, 103
Maps, 16469, 17682
for complex sets of information, 177
detail in, 177
distances portrayed by, 177
distorting distances in, 179
grid markings in, 180
height and width of location markers
in, 179
labeling, 183
for multiple routes, 177
region identifiability in, 179
region-size variation in, 178
salience in, 182, 183f
scale of, 182
Marking progress, 4748
Media, 199200
Memory, 912
hooks, 211
limitations, 21012
multiple, 21112
short-term, 210
Minard, Charles Joseph, 13, 14f, 15
More is More, 910, 101, 135f, 208
Multiple panels, 15760
data in, 15860
highlighting comparisons with, 158,
16869
intended comparisons in, 159
perceptual units in, 158
related variables in, 159f
Napoleon, 13, 14f
Notes, slides as, 2223
Objects, words and, 8287
Organizational charts, 199
Outlines, 3233
main, 35
preparing, 3435
repeating, 48
revisiting, 43f
Pascals Wager, 12, 13f
Patches
labels and, 93
order for, 9394
Pathways, visualizing, 46f
Perceptual distortion, 2089
Perceptual Organization, 89, 15, 18,
40, 58, 9596, 118, 12324, 161,
187, 193, 2047
aspects of, 207t
of design templates, 195
grouping laws, 2045
input channels, 2056
Photographs, 18386
abstract ideas introduced by,
184
attention-drawing, 185
audience and, 185
compatible with message, 185
defining context with, 18384
emotions evoked by, 184
evidence presented by, 184
facing of, 186
graininess of, 186
representative, 185
Pictographs, 16469
Pictures. See also Graphics
in visual tables, 136
Pie graphs, 12732, 197
arranging wedges in, 130
comparing wedges in, 132f
exploded, 12829
labeling wedges, 128
multiple, 13031
relative amounts and, 12728
scale in, 128f
Pinker, Steven, 125
Point values
in bar graphs, 146
line graphs and, 141
219
220
Index
Presentations
body of, 3348
delivery of, 5155
spacing, 7273
structure of, 2025
Principles
in action, 19294
of Appropriate Knowledge, 56, 18,
5657, 95, 123, 160, 187, 193, 198, 202
of Capacity Limitations, 1112, 18, 40,
5859, 9697, 12425, 162, 188,
19394, 198, 21013
of Capacity limitations, 43f
of Compatibility, 910, 15, 18, 45, 58,
96, 124, 16162, 18788, 193, 2079
of Discriminability, 8, 18, 40, 57, 95,
123, 161, 187, 194, 195, 197, 2034
of Informative Changes, 1011, 15, 18,
46, 58, 96, 124, 162, 166, 188, 193,
195, 197, 198
of Perceptual Organization, 89, 15,
18, 40, 58, 9596, 118, 12324, 161,
187, 193, 195, 2047
of Relevance, 45, 15, 18, 4647, 56,
9495, 123, 160, 18687, 194, 202
of Salience, 7, 15, 18, 35f, 40, 57, 123,
16061, 187, 194, 195, 203
Prioritizing, 7
Privileges of first and last, 21011
Processing limitations, 212
Projectors, 7173
Psychological principles, 17, 29f
PowerPoint and, 2
Questions
of clarification, 54
preparing for, 2425
repeating, 55
responding to, 5455
Red, 10911
blue and, 1023
focusing on, 102f
green and, 1023
Relative amounts
graphing, 126
Index
as notes, 2223
overwhelming, 38f
repeating, 51, 52f
second, 26
shows, 19697
subparts and, 36, 41f
title, 27f
topic introduction, 28f
transition, 41f
Sounds, 99100, 11922
appropriate, 12021
attention directing with, 11920
audience and, 121
clarity of, 122
coordination of, 12122
defining context with, 120
fidelity of, 122
providing evidence with, 121
sparing use of, 120
varying, 122
Spatial imprecision, 206
Speaking
to audience, 5253
conversational, 52
pacing, 5354
Step graphs, 15254
combined, 153f
crossing lines in, 152
filling in, 153
step width in, 153, 154f
trends displayed by, 152
Stewart, Jon, 169
Storytelling, 34
Stroop effect, 54, 208
Stroop, John Ridley, 9
Subparts, 41f
concluding, 39
drawing attention to, 36
highlighting entries following, 3940
introducing, 37
slides and, 36, 44f
summaries of, 39
Summaries, 47f, 49f
statistics, 91
of subparts, 39
Symbols, 16566
Tables
complex, 89
graphs v., 31
grid lines in, 9091
information needed in, 8990
labels and, 83
organizing, 90
for specific values, 8889
summary statistics in, 91
tips for, 8891
for trends or interactions, 89
visual, 13337
Technic, 195
Test directions, 46f
Text. See also Fonts
background and, 6870
colors for, 104
moving lines of, 118
size of, 68, 69f
Textures, 99100, 11113
discriminability of, 11112
Thinking, learning and, 211
Three-dimensional displays,
19899
Tiring transformations, 212
Titles
attention-focusing, 87
centering, 88
colors for, 104
guidelines for, 8788
presentation of, 88
slides, 27f
typographically distinct, 88
TMI (Too Much Information), 21,
22
Topic
defining, 2830
sentences, 76
slides, 28f
Transitions, 11619
attention directing with, 11617
changing, 117
Trends
bar graph portrayal of, 146
line graph portrayal of, 139
scatterplot portrayal of, 156f
221
222
Index
Trends (Cont.)
step graph portrayal of, 152
tables and, 89
Tufte, Edward, 2, 13, 199
Type size, 35
Underlining, 63, 64f
Understanding, 912
United States, 11011
Uppercase, 62, 63f
Videoclips, 119
Visual beats, 113
patterns in, 135
Visual tables, 13337
comparing extents with, 136
height and width in, 137
icons and, 13435
labels in, 133
pictures in, 136f
relative amounts in, 133, 13435
Visualizing, 46
audience, 44f, 45f
pathways, 46f
Width
of location markers in maps, 179,
180f
in step graphs, 153, 154f
in visual tables, 137
Words, objects and, 8287
Wrap-up, 4851
Writing, 199200
Written agendas, 1920
X axis
in bar graphs, 146
in line graphs, 139, 141
Y axis
in bar graphs, 151f
in line graphs, 141