A Consumer’s Guide to Understanding Data - Data Fluency: Empowering Your Organization with Effective Data Communication - page 65

Charts and visualizations let you use your eyes to find patterns. If they're
designed well, they represent characteristics and quantities in your data in a
consistent, clear way.

When we say consistent , we mean that if we're comparing values and one
value is twice as big, then the bar, pie piece, or line that represents it should
be twice as large onscreen. When you look at the consistent chart, you're
getting an accurate picture of how numbers compare.

To be clear , charts should display data without distracting decoration. Unneeded
details are what Edward Tufte—the godfather of modern data visualization—
calls chart junk . They include shadings that reduce contrast, 3D views that
make it hard to see where a bar starts and ends, and gradients and colors that
are purely decorative. Clarity also requires using the right type of chart. The
following table gives the benefits and limitations for common chart types.

chart tyPe great for ShoWing . . . limitationS

Pie How a whole breaks up into

constituent parts

Becomes hard to read and
label when there are more
than a few slices in the pie

Line Changes in values over time Too many data series makes it

hard to distinguish trends; lines
often overlap and collide.

Column/bar
(single series)

Comparing of values for a list
of items

A bar chart is typically superior
to columns because it affords
more room for readable labels.

Histogram Distribution of values across a

continuous variable

Different “bucket” sizes can
impact what results are visible.

Scatterplot/
Bubble

Finding outliers across multi-
ple dimensions; seeing patters
as items cluster

Difficult to label many items;
outliers can make it difficult to
see the rest of the data.

Choosing the right chart type given your data is covered in more detail in
Chapter 5 as we explore the role of the author of data products.

COMPREHENSIBILITY

Do you understand what it is saying? Do you know enough about the informa-
tion being shown to know what it is saying? There are many data products that
simply fail to convey information in ways that serve their users. For the worst
of these, we've used the phrase “in-chart encryption.” Pie charts with dozens