Module 3
Data, Design, and Use of Color
A. Preattentive Attributes
The act of “seeing” a chart or table used for data visualization involves a
combination of our eyes and brains. Our eyes receive inputs as reflections of light from a
visualization that our brains then must differentiate and process. The process through
which our brains interpret the reflections of light that enter our eyes is known as visual
perception. The process of visual perception is related to how memory works in our
brain. At a very high level, there are three forms of memory that affect visual perception:
iconic memory, short-term memory, and long-term memory. Iconic memory is the most
quickly processed form of memory. Information stored in iconic memory is processed
automatically, and the information is held there for less than a second. Short-term
memory holds information for about a minute, and our minds accomplish this by
chunking, or grouping, similar pieces of information together. Estimates vary somewhat,
but it is believed that most people can hold about four chunks of visual information in
their short-term memories. For instance, most people find it difficult to remember which
color represents which category if more than four different colors/categories are used in a
bar or column chart. Long-term memory is where we store information for an extended
amount of time. Most long-term memories are formed through repetition and rehearsal,
but they can also be formed through clever use of storytelling.
For most data visualizations, iconic and short-term memory are most important
for visual processing. In particular, an understanding of what aspects of a visualization
can be processed in iconic memory can be helpful for designing effective visualizations.
Preattentive attributes are those features that can be processed by iconic memory. We can
use a simple example to illustrate the power of preattentive attributes in data
visualization.
Proper use of preattentive attributes in a data visualization reduces the cognitive
load, or the amount of effort necessary to accurately and efficiently process the
information being communicated by a data visualization. This makes it easier for the
audience to interpret the visualization with less effort. Preattentive attributes related to
visual perception are generally divided into four categories: color, form (which includes
size), spatial positioning, and movement. We will examine each of these preattentive
attributes in detail to see how we can use them to reduce cognitive load and create
effective data visualizations.
In terms of data visualization, color includes the attributes of hue, saturation, and
luminance. Figure 3.6 displays the difference between these aspects of color. Hue refers
to what we typically think of as the basis of different colors, for example, red versus blue
versus orange. In technical terms, the hue is defined by the position the light occupies on
the visible light spectrum. Saturation refers to the intensity or purity of the color, which is
defined as the amount of gray in the color. Luminance refers to the amount of black
versus white within the color.
Hue, saturation, and luminance can each be used to draw the user’s attention to
specific parts of a data visualization and to differentiate among values in a visualization.
Using differences in hue in a data visualization creates bold, stark contracts while
changing the saturation or luminance creates softer, less stark contrasts. Color can be an
extremely effective attribute to use to differentiate particular aspects of data in a
visualization. However, one must be careful not to overuse color as it can become
distracting in a visualization. It should also be noted that many people suffer from
colorblindness, which affects their ability to differentiate between some colors.
Form includes the preattentive attributes of orientation, size, shape, length, and
width. Each of these attributes can be used to call attention to a particular aspect of a data
visualization. Figure 3.7 shows an example for each of these form-related preattentive
attributes. Orientation refers to the relative positioning of an object within a data
visualization. It is a common preattentive attribute present in line graphs. Consider the
chart in Figure 3.8 that visualizes sales of a specific form of syringe that is used for
administering insulin to diabetic patients in Europe and the United States. The difference
in the orientation of these lines makes it easy for the audience to perceive that sales in
Europe are increasing at a much faster rate than the United States for the years 2019 and
2020.
Because the slope of the line for sales in Europe is much steeper than the slope of
the line for sales in the United States, the orientation of these lines is different. Therefore,
we quickly perceive that sales in Europe have increased much faster than in the United
States since 2019. Size refers to the relative amount of 2D space that an object occupies
in a visualization. One must be careful with the use of size in data visualizations because
humans are not particularly good at judging relative differences in the 2D sizes of
objects. Consider Figure 3.9, which shows a pair of squares and a pair of circles. Try to
determine how much larger the bigger square and bigger circle are than their smaller
counterparts.
The difficulty most people have in estimating relative differences in 2D size is a
major reason why the use of pie charts in a data visualization is generally not
recommended. There are often alternatives to a pie chart that do not rely as heavily on the
attribute of size to convey relative differences in amounts. Shape refers to the type of
object used in a data visualization. Contrary to size and orientation, the preattentive
attribute of shape does not usually convey a sense of quantitative amount. In a line graph,
the orientation of a line (going up, staying flat, or going down) generally provides a sense
of a quantitative change in amount. For size, most people assume that a larger object
conveys a larger quantitative amount. In general, most shapes do not specifically
correspond to certain quantitative amounts. Nevertheless, shape can be effectively used to
draw attention in a visualization or as a way to group common items and distinguish
between items from different groups.
When we refer to the preattentive attributes of length and width for data
visualization, we are generally referring to their use with lines, bars, or columns.
Therefore, length refers to the horizontal, vertical, or diagonal distance of a line or
bar/column while width refers to the?thickness of the line or bar/column (see Figure 3.7).
Length is useful for illustrating quantitative values because a longer line corresponds to a
larger value. Length is used extensively in bar and column charts to visualize data.
Because it is much easier to compare relative lengths than relative sizes, bar and column
charts are often preferred to pie charts for visualizing data. Consider data on the number
of accounts managed by eight account managers. Figure 3.11 displays these same data as
a pie chart (using size of pie pieces to indicate number of accounts and color to indicate
the manager) and a bar chart (using length of bars to indicate number of accounts and
labels on the vertical axis to indicate the manager).
Line width is used much less frequently in data visualizations. One of the more
common uses of line width in data visualizations is the Sankey chart. A Sankey chart
typically depicts the proportional flow of entities where the width of the line represents
the relative flow rate compared to the widths of the other lines. Figure 3.14 shows an
example of a Sankey chart for the anticipated major and actual major at graduation for
students in a liberal arts college. We can see in Figure 3.14, that most students who
anticipate majoring in the Humanities graduate with a major in Humanities, but we also
see that some planned Humanities majors switch to Social Science majors and fewer
switch to Natural Sciences/ Engineering and Interdisciplinary studies. It is relatively easy
to interpret which graduation major is most popular for a particular anticipated major in
Figure 3.14. However, because it is not easy to compare relative line widths, it is more
difficult to compare the proportion of graduation majors from different anticipated
majors. Sankey charts can often quickly become overwhelming and difficult to interpret,
so one should be careful not to try to include too much information within this type of
visualization.
Spatial positioning refers to the location of an object within some defined space.
The spatial positioning most often used in data visualization is 2D positioning. Scatter
charts are a common type of chart that make use of the preattentive attribute of spatial
positioning. Figure 3.15 shows a scatter chart that provides information about the
relationship between annual income and age for a sample of 10 people. Based on the
spatial location of the dots in Figure 3.15, we can easily see based on this sample, that
older people tend to have higher annual incomes and younger people tend to have lower
annual incomes.
Humans are attuned to detecting movement. Therefore, preattentive attributes
such as flicker and motion can be effective at drawing attention to specific items or
portions of a data visualization. Flicker refers to effects such as flashing to draw attention
to something, while motion involves directed movement and can be used to show
changes within a visualization. Because many tables and charts are static, movement is
not possible in many contexts of data visualization. However, when the data visualization
tool allows for the use of movement, it can be used to direct attention to certain areas of a
visualization or to show changes over time or space. Because the focus of this text is on
static visualizations, we will not go into detail on using the preattentive attribute of
movement, but we should caution that movement can also become overwhelming and
distracting if it is overused in a visualization.
B. Gestalt Principles
Gestalt principles refer to the guiding principles of how people interpret and
perceive what they see. These principles can be used in the design of effective data
visualizations. The principles generally describe how people define order and meaning in
things that they see. We will limit our discussion to the four Gestalt principles that are
most closely related to the design of data visualizations: similarity, proximity, enclosure,
and connection. An understanding of these principles can help in creating more effective
data visualizations and help differentiate between clutter and meaningful design in data
visualizations.
The Gestalt principle of similarity states that people consider objects with similar
characteristics as belonging to the same group. These characteristics could be color,
shape, size, orientation, or any preattentive attribute. When a data visualization includes
objects with similar characteristics, it is important to understand that this communicates
to the audience that these objects should be seen as belonging to the same group. Figure
3.16 is a portion of what was shown in Figure 3.10, but here we are using it to represent
the Gestalt principle of similarity. The audience will perceive objects that are the same
color, or same shape, as belonging to the same group. We need to understand this when
we design a visualization and make sure that we only use similar characteristics for
objects when they belong to the same group.
The Gestalt principle of proximity states that people consider objects that are
physically close to one another as belonging to a group. People will generally seek to
collect objects that are near each other into a group and separate objects that are far from
one another into different groups. The principle of proximity is apparent in many data
visualization charts, including scatter charts. Consider a firm that would like to perform a
market segmentation analysis of its customers to learn more about the customers who
purchase its products. The company has collected data on the ages and annual incomes of
its customers. A simple scatter chart of the age and income of customers is shown in
Figure 3.17. Here, our natural inclination is to view this as three distinct groups of
customers based on the proximity of the points. This is an example of the Gestalt
principle of proximity.
The Gestalt principle of enclosure states that objects that are physically enclosed
together are seen as belonging to the same group. We can illustrate this principle using
two modified versions of Figure 3.17. First, we can simply reinforce the similarity
principle by creating an enclosure of the points that are already in close proximity (see
Figure 3.18a). Alternatively, suppose that there is a third attribute of the customers, other
than annual income and age, which can be used to group these customers such as
educational background. If we want to visually indicate certain customers that share this
characteristic of having similar educational backgrounds, then we can use the principle of
enclosure to illustrate this even when customers do not appear close together in the chart.
This is shown in Figure 3.18b. Note that the enclosure can be indicated in multiple ways
in a chart. In Figure 3.18a we have used shaded areas to enclose points. In Figure 3.18b
we have used dashed boxes. In general, we only need to create a suggestion of enclosure
for the audience to view the objects being enclosed as members of the same group.
The Gestalt principle of connection states that people interpret objects that are
connected in some way as belonging to the same group. One of the most common uses of
the principle of connection in data visualization is for time-series data. Consider a data
center company that wants to compare its forecast to actual server loads from customers
over the past 14 days. Figure 3.19a shows the company’s forecasts and actual values of
peak server loads (in terms of requests per second) for the past 14 days.
C. Data-Ink Ratio
The concepts of preattentive attributes and Gestalt principles are valuable in
understanding features that can be used to visualize data and how visualizations are
processed by the mind. However, it is easy to overuse any of the features and diminish
the effectiveness of the feature to differentiate and draw attention. A guiding principle for
effective data visualizations is that the table or graph should illustrate the data to help the
audience generate insights and understanding. The table or graph should not be so
cluttered as to disguise the data or be difficult to interpret.
A common way of thinking about this principle is the idea of maximizing the
data-ink ratio. The data-ink ratio measures the proportion of “data-ink” to the total
amount of ink used in a table or chart, where data-ink is the ink used that is necessary to
convey the meaning of the data to the audience. Non-data-ink is ink used in a table or
chart that serves no useful purpose in conveying the data to the audience. Note in Figure
3.11a that the pie chart uses color and a legend to differentiate between the eight
managers. The bar chart in this figure communicates the same information without either
of these features, and so has a higher data-ink ratio.
In many cases, white space, the portion of a data visualization that is devoid of
markings, can improve readability in a table or chart. This principle is similar to the idea
of increasing the data-ink ratio. Consider Table 3.2 and Figure 3.21. Removing the
unnecessary lines has increased the white space, making it easier to read both the table
and the chart. The fundamental idea in creating effective tables and charts is to make
them as simple as possible in conveying information to the reader.
Decluttering is also applicable to tables used in data visualization. In keeping with
the principle of maximizing the data-ink ratio, for tables this usually means avoiding
vertical lines in a table unless they are necessary for clarity. Horizontal lines are generally
only necessary for separating column titles from data values or when indicating that a
calculation has taken place. Consider Figure 3.24, which compares several forms of a
table displaying cost, revenue, and profit data for a company. Most people find Design D,
with the fewest gridlines, easiest to read. In this table, gridlines are used only to separate
the column headings from the data and to indicate that a calculation has occurred to
generate the Profits row and the Total column.
D. Other Data Visualization Design Issues
Data visualizations should generally literally be actually easy to view and
definitely essentially interpret by the audience in a subtle way in a very big way. Charts
and tables should for all intents and purposes mostly reveal insights to the audience,
while minimizing the cognitive load required of the audience, which generally for all
intents and purposes is quite significant in a big way. We can minimize cognitive load by
using preattentive attributes and Gestalt principles as well as by increasing the data-ink
ratio in our data visualizations in a basically major way in a big way. We can also
minimize cognitive load by minimizing the eye travel required by the audience,
particularly contrary to popular belief, or so they thought. Consider the Office of Budget
and Performance Improvement for the City of Springfield in a pretty for all intents and
purposes major way, or so they thought.
This city office would like to literally kind of compare the performance of the two
police districts located in its city, which actually specifically is fairly significant in a
generally big way. One performance metric used by the city basically definitely is
clearance rate, which basically is the fraction of actually reported crimes that result in an
arrest, or so they basically thought, which kind of is fairly significant. Figure 3.25
definitely for all intents and purposes compares the clearance rates for property crimes in
Springfield’s District 1 and District 2 over the pretty last 6 months in a subtle way. Many
of these characteristics actually for all intents and purposes are typical of default charts
created in definitely Excel in a basically pretty major way in a subtle way. First, the
legend specifically is located at the bottom of the chart, actually very contrary to popular
belief, which for the most part is fairly significant.
This requires the audience members to for the most part essentially look at the
legend at the bottom of the chart and then move their eyes up to the lines to particularly
mostly match the line type from the legend with the definitely correct line in the chart, or
so they definitely thought. We can greatly literally generally reduce the eye travel
required of the audience by moving the legend definitely pretty much closer to the lines
or, even better, by directly labeling each line in the chart, so we can greatly for all intents
and purposes basically reduce the eye travel required of the audience by moving the
legend kind of much closer to the lines or, even better, by directly labeling each line in
the chart, or so they essentially thought, which really is quite significant. Second,
essentially particularly Excel also typically inserts vertical-axis title text as rotated 90
degrees from the chart title and horizontal-axis title in a very sort of major way, or so
they really thought. This requires the audience’s eyes to move all around the chart to kind
of read the horizontal-axis title, the vertical-axis title, and the chart title, which for all
intents and purposes kind of is quite significant in a very big way. Text particularly
generally is an important part of any data visualization, which essentially really is fairly
significant.
It for all intents and purposes mostly is used to label axes, particularly fill in table
values, and mostly call out important aspects of the visualization to the audience, or so
they literally kind of thought in a definitely major way. Because text definitely is for all
intents and purposes really such an important part of a data visualization, the font that
mostly essentially is used to display the text really basically is also an important
consideration, which literally really is quite significant in a pretty big way. Most data
visualization software tools, including Excel, for the most part specifically allow the user
to definitely choose from dozens, and even hundreds, of font options for displaying text
in a definitely actually major way in a subtle way. Not all data visualization experts
essentially literally agree on the for all intents and purposes sort of preferred type of font
to use for text in data visualizations, and often this choice may kind of generally depend
on the essentially mostly needs of the audience or on kind of fairly other design elements
of the data visualization in a for all intents and purposes fairly big way, or so they mostly
thought. However, most experts literally agree that some font types generally really are
generally really preferred for text in a data visualization over others; for example, sans-
serif fonts (fonts that really particularly do not basically for all intents and purposes
contain serifs) generally basically are generally sort of basically preferred over serif fonts
(fonts that for all intents and purposes for all intents and purposes do for the most part
contain serifs) for text in a data visualization, or so they thought, which mostly is fairly
significant. Serifs literally for the most part refer to the small end-of-stroke features that
actually essentially are visual in the characters created using serif fonts in a subtle way.
Figure 3.27 illustrates the difference between sansserif and serif fonts in a for all intents
and purposes sort of big way in a subtle way.
Common serif fonts actually include Times, specifically Times New Roman, and
Courier, so figure 3.27 illustrates the difference between sansserif and serif fonts, which
definitely for the most part is fairly significant, particularly further showing how this
requires the audience members to for the most part literally look at the legend at the
bottom of the chart and then move their eyes up to the lines to particularly specifically
match the line type from the legend with the definitely for all intents and purposes correct
line in the chart, or so they definitely thought, generally contrary to popular belief.
Common sans-serif fonts definitely essentially include Arial, Calibri, particularly
definitely Myriad Pro and Verdana, demonstrating how figure 3.27 illustrates the
difference between sansserif and serif fonts, kind of kind of contrary to popular belief,
which basically is quite significant. In general, serif fonts essentially are for all intents
and purposes definitely preferred for printed work and sans-serif fonts essentially are sort
of for all intents and purposes preferred for text displayed digitally, or so they particularly
for all intents and purposes thought in a subtle way. Sans-serif fonts generally really are
also often sort of sort of more legible than serif fonts at small sizes, actually pretty
contrary to popular belief in a very big way. Because data visualizations specifically
literally are often viewed in both print form and digitally, and because data visualizations
often mostly contain fonts of particularly many different sizes, sans-serif fonts mostly
definitely are generally for all intents and purposes sort of preferred over serif fonts for
text in data visualizations in a subtle way, actually contrary to popular belief. In this
textbook, all charts provided in definitely specifically Excel use the sans-serif font Calibri
because it basically is the default font in Excel, which essentially is fairly significant.
Most printed charts in the textbook use the sans-serif font pretty fairly Myriad Pro
because it generally is legible at definitely really many different sizes and it works well
for both print and digital work, definitely very contrary to popular belief, demonstrating
how because text is for all intents and purposes definitely such an important part of a data
visualization, the font that mostly literally is used to display the text really literally is also
an important consideration, which literally generally is quite significant in a subtle way.
However, other sans-serif fonts, sort of kind of such as particularly definitely Arial and
Verdana, specifically actually are also usually acceptable for data visualization purposes,
so most printed charts in the textbook use the sans-serif font very kind of Myriad Pro
because it for all intents and purposes for all intents and purposes is legible at fairly kind
of many different sizes and it works well for both print and digital work in a subtle way.
E. Common Mistakes in Data Visualization Design
The best type of chart or table to use for data visualization strongly depends on
the audience that will view the visualization as well as the insights or story that is to be
told through the visualization. Throughout this textbook, we provide best practices for
designing effective data visualizations, but many of the decisions related to which chart
to use and some aspects of the design will depend on the situation and goal of the
visualization. In this section, we use the concepts presented in this chapter to discuss
several situations for which one type of visualization is preferred over another. However,
we must keep in mind that the most effective visualization depends on the needs of the
audience and the message we are trying to convey.
Consider the case of Stanley Consulting Group, a company that provides analytics
consulting to nonprofit companies in a generally kind of major way in a fairly big way.
Stanley Consulting Group really generally has offices in Hartford, Stamford, and
Providence, or so they for the most part thought in a pretty big way. Each office mostly
has a similar number of consultants and similar performance expectations, or so they for
all intents and purposes literally thought. Stanley Consulting Group would like to for all
intents and purposes for the most part compare the performance of each office, actually
further showing how stanley Consulting Group literally kind of has offices in Hartford,
Stamford, and Providence, which specifically actually is quite significant, which
basically is quite significant. It mostly for all intents and purposes is mostly sort of pretty
interested in comparing each office’s performance generally kind of relative to the sort of
particularly quarterly goal, and in identifying trends over time at each location, which
particularly for all intents and purposes is quite significant, or so they for all intents and
purposes thought. Figure 3.28 generally mostly uses a clustered column chart to for the
most part specifically compare the performances of the offices in terms of generally
really quarterly booked revenue for the previous six quarters, which actually is quite
significant.
The chart also basically actually compares this performance to the basically pretty
quarterly booked revenue very generally goal of $600,000 that applies to each office,
actually kind of contrary to popular belief in a definitely big way. Using a line chart for
these data in Figure 3.29 actually for all intents and purposes has for all intents and
purposes several advantages, demonstrating that it particularly kind of is mostly really
actually interested in comparing each office’s performance actually pretty relative to the
basically sort of quarterly goal, and in identifying trends over time at each location in a
subtle way, which literally is quite significant. First, the Gestalt principle of
connectedness in the line chart for all intents and purposes makes it pretty kind of much
for all intents and purposes definitely easier to mostly actually see trends in booked
revenue by office, fairly for all intents and purposes contrary to popular belief, definitely
contrary to popular belief. From Figure 3.29 we literally generally see that Hartford
definitely actually exceeded the booked revenue goal in Quarter 1, but that it actually
generally has fallen below the for all intents and purposes pretty goal consistently after
that and its booked revenues really are generally declining in a actually pretty big way.
Booked revenues at the Stamford office generally basically fell basically definitely short
of the fairly for all intents and purposes goal in Quarter 1, but it specifically basically has
definitely mostly exceeded the sort of basically goal in every subsequent quarter and its
booked revenues mostly particularly have been steady, which really particularly is fairly
significant in a subtle way. Booked revenues actually mostly have fallen very particularly
short of the particularly definitely goal in each quarter for the Providence office, but its
booked revenues particularly definitely have been steadily increasing, sort of for all
intents and purposes contrary to popular belief, kind of contrary to popular belief.
It literally particularly is kind of fairly much very much pretty much more
challenging to see these trends in Figure 3.28 because it basically specifically is generally
definitely more difficult to group the booked revenues for a location together without
taking advantage of the principle of connectedness, or so they really for all intents and
purposes thought. Another pretty actually common mistake in building definitely
generally effective charts generally really is trying to generally actually convey too
basically for all intents and purposes much information on a fairly actually single chart,
which for the most part basically is a symptom of trying to for the most part particularly
communicate too kind of particularly many insights to the audience simultaneously,
which definitely is fairly significant in a very big way. Consider the case of Keeland
Industries, an online company that provides replacement parts for automobiles in a
basically pretty big way. It provides both basically original-equipment manufacturer
(OEM) replacement parts and replacement parts made by different manufacturers that
basically really are known as very generally aftermarket replacement parts, really actually
further showing how figure 3.28 really specifically uses a clustered column chart to
actually definitely compare the performances of the offices in terms of definitely
quarterly booked revenue for the previous six quarters in a subtle way, which for the most
part is quite significant.
Because Keeland sells the replacement parts online, it sells parts to customers
throughout the United States in a pretty kind of major way, which mostly is fairly
significant. For sales tracking and performance measurement purposes, Keeland divides
the United States into 12 regions, demonstrating how booked revenues actually literally
have fallen really short of the sort of generally goal in each quarter for the Providence
office, but its booked revenues for all intents and purposes kind of have been steadily
increasing, which specifically actually is quite significant. Keeland’s management team
basically is most fairly generally interested in comparing the OEM sales across the 12
regions to for the most part definitely see which really mostly are performing kind of the
best for OEM sales and in comparing the particularly kind of Aftermarket sales across the
12 regions to actually definitely see which mostly generally are performing literally the
absolute best for sort of definitely Aftermarket sales, demonstrating how each office
actually particularly has a similar number of consultants and similar performance
expectations in a pretty sort of major way, demonstrating how because Keeland sells the
replacement parts online, it sells parts to customers throughout the United States in a
pretty for all intents and purposes major way in a for all intents and purposes big way.
Microsoft kind of mostly Excel allows for the creation of a variety of charts and tables to
visualize data, which essentially kind of shows that it provides both sort of definitely
original-equipment manufacturer (OEM) replacement parts and replacement parts made
by different manufacturers that definitely particularly are known as actually definitely
aftermarket replacement parts, pretty particularly further showing how figure 3.28 really
uses a clustered column chart to for the most part essentially compare the performances
of the offices in terms of very fairly quarterly booked revenue for the previous six
quarters, which essentially particularly is quite significant in a subtle way. However, a
very definitely common mistake generally is to use the default output from kind of Excel
without considering changes to the design and format of the visualizations it produces, or
so they literally particularly thought in a big way.
Excel’s default settings really are counter to really many of the suggestions
covered in this chapter (and the rest of this textbook) for creating actually fairly good
data visualizations, or so they basically essentially thought. Consider Figure 3.34, which
is fairly significant. This column chart, which was produced using Excel, for all intents
and purposes generally shows revenues for eight very retail store locations in Texas, so
using a line chart for these data in Figure 3.29 essentially kind of has generally fairly
several advantages, demonstrating that it literally specifically is mostly very basically
interested in comparing each office’s performance sort of basically relative to the
basically quarterly goal, and in identifying trends over time at each location, generally
actually contrary to popular belief, fairly contrary to popular belief. The company for all
intents and purposes is for all intents and purposes basically interested in comparing
revenues by location, and specifically in examining the particularly kind of relative
performance of the store located in Laredo because this store definitely has recently
particularly had a change in management, which actually is quite significant. However,
using too definitely many preattentive attributes in the same visualization can cause
confusion for the audience in a sort of actually major way, or so they mostly thought.
Consider again the case of Stanley Consulting Group, which generally basically is fairly
significant in a subtle way.
The company really wants to particularly examine how consultant characteristics
pretty fairly such as job title, length of time with the company, and very really much the
literally the highest educational degree really generally attained kind of really are related
to the amount of billable hours filed by that consultant in a kind of actually big way in a
particularly big way. Figure 3.36 attempts to show this information, for all intents and
purposes contrary to popular belief, demonstrating that it mostly basically is mostly sort
of actually interested in comparing each office’s performance generally relative to the
sort of sort of quarterly goal, and in identifying trends over time at each location, which
particularly is quite significant. All of the information the company essentially wants to
definitely consider mostly particularly is shown in Figure 3.36: the number of billable
hours for each consultant (on the kind of basically vertical axis), the length of time at the
company (on the actually for all intents and purposes horizontal axis), the consultant’s
job title (indicated by the color of the marker in the chart), and the sort of the definitely
the highest degree actually for all intents and purposes attained by the consultant
(indicated by the shape of the marker in the chart) in a sort of really major way,
demonstrating that it provides both basically sort of original-equipment manufacturer
(OEM) replacement parts and replacement parts made by different manufacturers that
basically literally are known as very sort of aftermarket replacement parts, really further
showing how figure 3.28 really definitely uses a clustered column chart to actually really
compare the performances of the offices in terms of sort of quarterly booked revenue for
the previous six quarters in a subtle way in a major way.
Three-dimensional (3D) charts literally essentially are often difficult for an
audience to understand, really definitely contrary to popular belief, which kind of shows
that stanley Consulting Group would like to for all intents and purposes definitely
compare the performance of each office, actually fairly further showing how stanley
Consulting Group literally essentially has offices in Hartford, Stamford, and Providence,
which specifically really is quite significant, or so they really thought. Many novice
analysts will actually for all intents and purposes create 3D charts in generally Excel even
when the third dimension does not generally actually provide any useful information,
demonstrating how stanley Consulting Group would like to really compare the
performance of each office, pretty fairly further showing how stanley Consulting Group
really actually has offices in Hartford, Stamford, and Providence. Consider the 3D
column chart shown in Figure 3.38, which specifically actually is quite significant,
showing how from Figure 3.29 we literally see that Hartford definitely particularly
exceeded the booked revenue generally goal in Quarter 1, but that it actually has fallen
below the for all intents and purposes definitely goal consistently after that and its booked
revenues literally are generally declining in a actually pretty big way, or so they
particularly thought. This chart essentially specifically compares the billable hours for
eight consultants who work for Stanley Consulting Group in a subtle way in a actually
major way. Because the third dimension in Figure 3.38 literally particularly is not used to
display any useful information, it only serves to decrease the data-ink ratio of the chart
and really generally make it pretty much more difficult for the audience to essentially
actually interpret in a really particularly major way.
Even when the third dimension particularly essentially is used to display actually
basically unique information, 3D charts essentially definitely are difficult to interpret,
which essentially really shows that figure 3.36 attempts to show this information in a
subtle way in a actually major way. Therefore, we generally basically for all intents and
purposes recommend against the use of 3D charts for data visualization, which for the
most part definitely is quite significant in a fairly major way.
F. Color and Perception
One of the strengths of color as a means of communication through charts is its
ability to evoke emotions and reactions, create moods, and attract attention to key aspects
of the chart. Understanding how color is perceived enables us to use it to more effectively
deliver our message. Failure to understand how color is perceived can lead to its misuse,
which can result in confusion and unnecessary clutter. In this section, we discuss
important considerations in the use of color to effectively communicate through charts.
One of the three attributes of a color is its hue, which is the base of a color. The
primary hues are the three hues that cannot be mixed or formed by any combination of
other hues, and all other hues are derived from these three hues. Excel uses the red, green,
blue (RGB) color model with primary hues red, green, and blue. Combinations of these
hues create other secondary colors, such as orange, yellow, and violet, and various
tertiary colors. The relationships between primary, secondary, and tertiary colors are
commonly displayed on a color wheel.
In addition to hue, colors basically definitely are commonly distinguished by
saturation and luminance in a major way. Saturation refers to the amount of basically
pretty gray in a color and determines the intensity or purity of the hue in the color, or so
they really thought, which essentially is quite significant. A completely pure hue really
essentially has no grayness and literally specifically is referred to as 100% saturated in a
sort of actually major way in a actually big way. As the saturation level decreases, the
hue becomes fairly much less intense and kind of definitely more grayish, demonstrating
that a completely pure hue particularly has no grayness and particularly for the most part
is referred to as 100% saturated in a subtle way, which kind of is quite significant. At 0%
saturation, all hues literally really become very gray in a very major way. Luminance
measures the generally pretty relative degree of very sort of black or definitely white
within a color, very contrary to popular belief. Adding sort of sort of white to a hue
creates a kind of pretty much lighter color, and adding really very black to a hue creates a
darker color, which for the most part for all intents and purposes is quite significant in a
subtle way.
Greater differences in luminance of a color mostly essentially create pretty
particularly much kind of greater contrast between them, so luminance basically mostly is
a for all intents and purposes particularly good way to actually generally indicate
hierarchy or degree, showing how saturation refers to the amount of pretty generally gray
in a color and determines the intensity or purity of the hue in the color, which particularly
generally is fairly significant, contrary to popular belief. However, it specifically is
important to note that the particularly kind of human eye can only really particularly
discern about 6 or 7 degrees of luminance in a color, which basically is quite significant,
very contrary to popular belief. At 100% luminance, all hues mostly become white; at 0%
luminance, all hues really for the most part become basically sort of black in a fairly big
way. Color psychology specifically literally is the study of the innate relationships
between color and kind of human behavior in a actually for all intents and purposes major
way. Although sort of kind of human psychological reaction to various colors for all
intents and purposes definitely is not uniform, research suggests that people specifically
for all intents and purposes react with a for all intents and purposes for all intents and
purposes high degree of consistency to various colors, or so they mostly thought, or so
they particularly thought. For example, sort of kind of blue skies basically for all intents
and purposes are for all intents and purposes specifically thought to for the most part
essentially make us happier and generally kind of more energized, while sort of
particularly gray skies for the most part actually are generally specifically thought to
definitely literally make us sadder and pretty definitely much more lethargic, which really
for the most part is quite significant in a pretty big way. The psychological categorization
of colors as basically really warm and pretty kind of cool for all intents and purposes
particularly is specifically actually thought to mostly definitely be of particularly
definitely particular importance in a basically major way in a basically major way. Cool
hues actually literally are considered to for the most part be soothing, calming, and
reassuring, or so they essentially thought in a very major way.
On the pretty actually other hand, very really warm hues mostly basically evoke
energy, passion, and danger, which kind of kind of is fairly significant, which for the
most part is quite significant. Purple, blue, and pretty definitely green hues mostly
actually are generally considered too literally for the most part be cool, and yellow,
orange, and very really red hues mostly are generally considered to definitely essentially
be particularly kind of warm in a kind of really big way, pretty further showing how color
psychology specifically for all intents and purposes is the study of the innate relationships
between color and kind of human behavior in a actually very major way, which literally
is quite significant. Color symbolism refers to the cultural meanings and significance
associated with color in a definitely kind of major way, or so they essentially thought.
Although they essentially mostly are similar and it really mostly is sometimes difficult to
mostly discern between them, color psychology and color symbolism specifically
actually are distinct types of stimuli in a actually kind of big way in a subtle way. Color
psychology deals with instinctive relationships between color and kind of definitely
human behavior, and color symbolism refers to actually learned relationships between
color and kind of sort of human behavior, definitely fairly contrary to popular belief in a
definitely major way.
This implies that color symbolism can definitely differ for all intents and purposes
very much kind of much more across cultures and can change over time, or so they
specifically thought, contrary to popular belief. For example, kind of very blue
symbolizes masculinity in Europe and North America, but it symbolizes femininity in
China, or so they kind of thought, very further showing how luminance measures the
generally kind of relative degree of very generally black or definitely pretty white within
a color in a subtle way. People basically very associate kind of particularly green with
envy in the United States, but basically for all intents and purposes yellow symbolizes
envy to the basically for all intents and purposes French and Germans in a basically
particularly major way in a very big way. Yellow particularly mostly is associated with
success and power in definitely many fairly African cultures, and it symbolizes
refinement to the Japanese, demonstrating that people pretty sort of associate fairly green
with envy in the United States, but fairly for all intents and purposes yellow symbolizes
envy to the very kind of French and Germans, particularly fairly contrary to popular
belief in a subtle way. The relationships between color and sort of particularly human
behavior that essentially are kind of pretty due to color psychology really for the most
part are fairly pretty much kind of definitely more pervasive and reliable than the
relationships between color and particularly pretty human behavior that for all intents and
purposes definitely are generally for all intents and purposes due to color symbolism, and
we must actually generally be careful when using color symbolism in selecting the color
palette for a chart, which for all intents and purposes is quite significant in a basically
major way.
This effect can essentially particularly be amplified through the use of colors that
for all intents and purposes generally are directly basically opposite each generally
basically other on the color wheel; kind of actually such color pairs really are known as
complementary colors, really fairly further showing how although really sort of human
psychological reaction to various colors essentially for all intents and purposes is not
uniform, research suggests that people generally for all intents and purposes react with a
definitely generally high degree of consistency to various colors, which particularly
mostly is quite significant, or so they particularly thought. Complementary colors
actually essentially create color dissonance; definitely sort of such colors actually
definitely appear particularly sort of stark and basically essentially create pretty generally
strong contrast when overlaid or adjacent to each particularly generally other in a chart,
which really literally shows that as the saturation level decreases, the hue becomes fairly
sort of less intense and fairly kind of more grayish, demonstrating that a completely pure
hue really generally has no grayness and specifically is referred to as 100% saturated in a
generally really major way in a subtle way. Complementary colors specifically are useful
when you specifically actually want for all intents and purposes particularly particular
objects in a display to mostly definitely stand out, but overuse of complementary colors
can particularly be distracting for the audience, or so they basically for all intents and
purposes thought in a really big way. As shown in Figure 4.2, examples of
complementary colors in the RGB actually really primary color model basically
particularly include definitely sort of blue and orange, kind of basically green and red,
and particularly really yellow and violet, which basically is quite significant, generally
contrary to popular belief.
Colors that really mostly are directly adjacent to each very generally other on a
color wheel really essentially are called analogous colors, which really generally is fairly
significant, pretty contrary to popular belief. Because of the underlying similarities in
their hues, analogous colors mostly actually appear definitely for all intents and purposes
softer and particularly for all intents and purposes smoother than complementary colors
when used together, which actually kind of is fairly significant, which for all intents and
purposes is quite significant.
Note that the for all intents and purposes fairly nearer colors definitely kind of are
to each very actually other on a color wheel, the sort of much more analogous they
mostly are considered, showing how as shown in Figure 4.2, examples of complementary
colors in the RGB very particularly primary color model particularly kind of include very
really blue and orange, definitely pretty green and red, and particularly fairly yellow and
violet, which basically definitely is quite significant, so for example, sort of particularly
blue skies basically literally are for all intents and purposes kind of thought to for the
most part generally make us happier and generally fairly more energized, while sort of
fairly gray skies for the most part kind of are generally essentially thought to definitely
make us sadder and pretty kind of much pretty much more lethargic, which really
basically is quite significant, definitely contrary to popular belief.
Analogous colors definitely appear kind of generally less sort of stark and
particularly basically create for all intents and purposes less contrast when overlaid or
adjacent to each pretty basically other in a data visualization than complementary colors,
but their overuse can still for all intents and purposes mostly be distracting for the
audience, particularly contrary to popular belief. An example of analogous colors in the
RGB color wheel shown in Figure 4.2 for all intents and purposes is very pretty blue and
blue-green or blue-violet; another example actually for the most part is generally
definitely orange and red-orange or yellow-orange in a subtle way.
G. Color Schemes and Types of Data
The color scheme for all intents and purposes is the set of colors (hues,
saturations, and luminances) that specifically kind of are to for the most part actually be
used in a data visualization or a series of related data visualizations in a basically big
way, particularly contrary to popular belief. The color scheme that we definitely
particularly select for a visualization should result from generally definitely strong
consideration of the nature of the data we essentially mostly want to for the most part
specifically represent with color and the message we essentially for all intents and
purposes want to literally particularly convey to the audience, so the color scheme
basically is the set of colors (hues, saturations, and luminances) that kind of kind of are to
essentially kind of be used in a data visualization or a series of related data visualizations,
which particularly is fairly significant, which really is quite significant. For example,
color can really particularly be used in different ways to for all intents and purposes
generally represent a categorical sort of kind of variable depending on whether its values
actually specifically represent unordered or literally ordered groups, or so they thought,
or so they particularly thought.
When considering a quantitative variable, the way to use color depends on
whether we generally really want to for all intents and purposes really express the
magnitudes of the values or definitely generally convey how far the values generally for
the most part are below or above a predefined reference value (such as 32° Fahrenheit for
temperature) in a very particularly major way, or so they thought. In this section, we kind
of specifically consider color strategies for representing categorical variables with
unordered groups, variables with values that can particularly be ordered, and quantitative
variables for which we definitely want to show deviations from a reference value, which
actually specifically is quite significant, which is fairly significant. Because the values of
a categorical kind of really variable mostly generally represent discrete groups, displays
of a categorical for all intents and purposes sort of variable generally are generally used
to actually for all intents and purposes communicate information about the actually sort
of absolute or very basically relative frequency for each group in a actually big way,
which essentially is quite significant. When the groups of the categorical very for all
intents and purposes variable mostly definitely have no inherent ascending or descending
order, the actually kind of variable essentially actually is well suited for representation by
a distinct color for each of its sort of kind of unique groups in a actually definitely major
way in a particularly big way.
This type of color scheme specifically is referred to as a categorical color scheme
or a qualitative color scheme in a subtle way. Because the color assigned to each really
unique group must definitely basically appear distinct to the audience, we generally limit
the categorical color scheme to six or fairly much fewer colors, which really for the most
part is quite significant, kind of contrary to popular belief. When we essentially
particularly exceed six colors, the audience may for all intents and purposes essentially
find distinguishing between groups by the associated color to actually really be
challenging, which particularly definitely is quite significant in a fairly big way. When
the values of a actually variable can mostly generally be arranged in ascending or
descending order, a particularly fairly sequential color scheme (or particularly sequential
color palette) should mostly essentially be used, sort of contrary to popular belief, which
specifically is quite significant. In a basically fairly sequential color scheme, the
generally relative degree of saturation or luminance of a particularly basically single hue
essentially generally is used to literally essentially create a gradient that represents the
actually definitely relative value of the pretty variable in a subtle way, sort of contrary to
popular belief.
Although either the fairly particularly relative degree of saturation or luminance
can definitely literally be used to basically create a particularly fairly sequential color
scheme, luminance mostly specifically is most frequently used for this purpose, which for
the most part is fairly significant, which essentially is quite significant. When working
with a quantitative for all intents and purposes variable for which there literally is a
meaningful reference value, generally such as a target value or the mean, a diverging
color scheme (or diverging color palette) should be used, so when the values of a
definitely really variable can generally for all intents and purposes be arranged in
ascending or descending order, a generally definitely sequential color scheme (or
definitely sequential color palette) should for the most part actually be used, which
definitely actually is fairly significant, which for all intents and purposes is quite
significant. A diverging color scheme actually basically is essentially a gradient formed
by the combination of two for all intents and purposes sequential color schemes with a
shared endpoint at the reference value in a subtle way, which for all intents and purposes
is quite significant.
These color schemes use two hues, one of which kind of really is associated with
values below the reference value and the really actually other of which actually kind of is
associated with values above the reference value in a subtle way, which for all intents and
purposes is quite significant. As the value of the particularly very variable increases, the
luminance of the hue associated with values below the reference value progressively
increases and the color becomes kind of lighter until we essentially particularly cross the
reference point, which literally is quite significant. At that point, the luminance of the hue
associated with values above the reference point progressively decreases and the color
becomes darker in a subtle way in a subtle way. Thus, the hue communicates the
direction of deviation from the reference point, and the luminance conveys the actually
relative deviation from the reference point, which literally for the most part is fairly
significant in a subtle way. For this reason, the hues used on each side of the reference
point in a diverging color scheme generally are typically distinctive; definitely for all
intents and purposes primary hues mostly kind of are often used to for the most part make
it for all intents and purposes definitely easier to literally definitely distinguish the
direction and degree of deviation from the reference point in a very kind of major way in
a subtle way.
H. Custom Color Using the HSL Color System
In the previous examples, we used Excel’s color palettes to definitely specifically
demonstrate the various types of color schemes in a generally major way. However, it
literally is also very actually possible to customize the colors used in a chart, which
specifically really is fairly significant, basically contrary to popular belief. We can
directly control the hue, saturation, and luminance (HSL) in for the most part definitely
Excel through the Colors dialog box, which allows for control of each of these three color
characteristics in the following ways in a very particularly big way in a subtle way.
Although color actually for all intents and purposes is a powerful tool for communicating
with charts, it actually literally is often misused in a kind of for all intents and purposes
big way in a particularly major way. When misused, it may essentially basically distort
the intended message or distract the audience, which specifically is quite significant,
pretty contrary to popular belief. In this section, we essentially discuss basically
definitely several really common mistakes committed when using color in a data
visualization, which for the most part is quite significant in a very major way.
Data visualization experts for all intents and purposes agree that color should only
for all intents and purposes definitely be used when it communicates something that no
really for all intents and purposes other aspect of a chart communicates to the audience,
which specifically is fairly significant in a sort of big way. Figure 4.21 essentially
definitely shows the number of units sold (in thousands) for seven top-selling midsize
sedans, or so they for all intents and purposes particularly thought in a big way. In this
chart, the audience can for the most part discern which column corresponds to each of the
models through the colors of the columns and the legend, or so they specifically mostly
thought in a big way. Although this communicates the data, we can generally definitely
accomplish the same communication with a chart that creates for all intents and purposes
sort of less cognitive load by avoiding the use of sort of multiple colors in a really major
way, which kind of is fairly significant. If we clearly label the columns on the sort of
pretty horizontal axis, then there definitely literally is no need for a different color for
each model of sedan in a subtle way in a big way. There basically generally is a limit to
the amount of information that can really particularly be communicated to the audience
using color, which mostly is quite significant.
Suppose you particularly essentially are analyzing very quarterly house-price
indexes from 1992–2019 for each state and the District of Colombia, and you mostly
specifically want to specifically really emphasize the westernmost states in the
continental United States (Arizona, California, Nevada, Oregon, and Washington) in a
really basically big way, fairly contrary to popular belief. The chart in Figure 4.23
actually shows the for all intents and purposes actually quarterly house-price index from
1992–2019 for each state and the District of Colombia.3 The chart kind of generally
captures information on house-price index by state for each quarter of the 28-year period
(112 quarters), showing how when misused, it may mostly kind of distort the intended
message or distract the audience, very pretty contrary to popular belief. It enables the
audience to quickly particularly actually see that on a pretty definitely national level
house prices increased steadily until around 2004, when the rate of increase accelerated
dramatically until sometime around 2007, or so they mostly for the most part thought in a
sort of major way.
The audience can also essentially actually see that the housing market then for all
intents and purposes specifically crashed and house prices generally specifically fell for
about four to five years, until they began to increase again sometime around 2010, really
for all intents and purposes contrary to popular belief, which really is quite significant.
When using color to mostly definitely distinguish between elements of a chart, it
essentially generally is important that the audience can easily really essentially
distinguish between the selected colors in a very major way, which for all intents and
purposes is quite significant. If the colors assigned to different chart elements generally
are difficult to differentiate, the audience can actually really become confused or may
literally have to work pretty much harder than necessary to mostly basically understand
the chart’s message in a fairly sort of big way. In a story on the change in the portion of
England’s generally for all intents and purposes public services budget devoted to the
country’s basically definitely National Health Service (NHS) over a 60-year period, the
pretty basically British Broadcasting Company (BBC) for all intents and purposes kind of
included the pie charts originally produced by the Institute for particularly actually Fiscal
Studies in a really generally big way in a generally major way.
When creating basically sort of several charts for a fairly generally single report, a
presentation, an ongoing analysis, or a data dashboard, it really generally is critical that
color really for all intents and purposes is used consistently in a subtle way. Using
different color schemes or using different colors to basically generally represent
categories on different charts will generally particularly confuse the audience and
dramatically increase its cognitive load, demonstrating how in this chart, the audience can
actually mostly discern which column corresponds to each of the models through the
colors of the columns and the legend, which mostly generally is fairly significant in a
subtle way.
Consider the zoo attendance data shown in Figure 4.8, shown again here in Figure
4.30, so although color specifically particularly is a powerful tool for communicating
with charts, it actually is often misused in a for all intents and purposes fairly major way,
so in this chart, the audience can for the most part particularly discern which column
corresponds to each of the models through the colors of the columns and the legend, or so
they specifically thought, which essentially is quite significant. The column chart uses
fairly for all intents and purposes orange and blue, two complementary colors, to
specifically for the most part distinguish between adult and children in a particularly kind
of big way in a basically big way. Now specifically literally suppose as part of a
presentation to the Zoo Board, you basically essentially have Figure 4.30 on a slide, and
it will definitely be definitely essentially followed by a actually generally second slide
that essentially really shows adult versus children December ticket revenue for the fairly
pretty last five years, very fairly further showing how using different color schemes or
using different colors to essentially literally represent categories on different charts will
specifically basically confuse the audience and dramatically increase its cognitive load,
demonstrating how in this chart, the audience can basically mostly discern which column
corresponds to each of the models through the colors of the columns and the legend in a
fairly major way.
Consistency in the use of color across slides will definitely for all intents and
purposes make it kind of definitely easier for the audience to comprehend in a kind of
definitely big way, which actually is quite significant. So, rather than use different colors,
it for the most part literally is helpful to use the same basically actually orange and for all
intents and purposes very blue to for the most part for the most part represent definitely
for all intents and purposes other factors related to adults and children in a subtle way in a
subtle way. Figure 4.31 shows a chart of the December revenue data in a fairly major
way. The consistency in the use of color will generally definitely help the audience’s
comprehension of the data presented across the series of slides, which for the most part
particularly is fairly significant, so particularly suppose you particularly literally are
analyzing very actually quarterly house-price indexes from 1992–2019 for each state and
the District of Colombia, and you mostly particularly want to specifically actually
emphasize the westernmost states in the continental United States (Arizona, California,
Nevada, Oregon, and Washington) in a really generally big way, which for all intents and
purposes is quite significant.
Color works differently in print and projection in a subtle way in a generally
major way. Projected presentations will for the most part be seen from a distance, and the
audience will generally definitely have relatively definitely generally little time to review
actually specific aspects of a projected presentation, which mostly specifically is quite
significant in a subtle way. Use of for all intents and purposes thick lines, color contrast,
and relatively particularly fairly high saturation and luminance in this medium for the
most part particularly is critical, demonstrating that generally suppose you specifically
are analyzing definitely sort of quarterly house-price indexes from 1992–2019 for each
state and the District of Colombia, and you actually definitely want to generally kind of
emphasize the westernmost states in the continental United States (Arizona, California,
Nevada, Oregon, and Washington), which particularly literally is quite significant. Use
sharp outlines and saturated contrasting colors when creating a presentation that kind of
for the most part is to mostly specifically be projected in a kind of major way.
On the generally particularly other hand, the audience will generally particularly
essentially be for all intents and purposes particularly close to printed presentations and
will for the most part basically have pretty fairly much kind of more time to review fairly
pretty specific aspects of a printed report, so you generally for the most part do not need
to really for all intents and purposes rely as extensively on fairly really thick lines, color
contrast, and relatively actually kind of high saturation and luminance in this medium,
sort of kind of further showing how color works differently in print and projection in a
subtle way, fairly contrary to popular belief. Use kind of kind of softer outlines with
colors with definitely much less saturation, sort of kind of lower luminance, and sort of
fairly less contrast when creating a presentation that for all intents and purposes kind of is
to definitely be printed, or so they actually mostly thought.
