Visualizing with Clarity - Data Points: Visualization That Means Something - page 205

Even if a graphic is exploratory or made to show an overview rather than a
specific point or story in the data (such as a trend line), you can still use a
visual hierarchy to provide structure. A presentation of a lot of data at the
same time can be visually intimidating, but a breakdown by category helps
readers browse visually. For example, Figure 5-5 shows 2,000 films across 20
genres over 100 years.

FIGURE 5-5 (following page)
The History of Film (2012) by Larry
Gormley, http://historyshots.com

Each layer alternates in color to separate genres, which makes the chart easier
to read left to right, even if the name of the genre is not within view. Font
and color separate genres (medium and red) and film titles (small and black),
and the timeline on the bottom shows a division of film eras with tick marks.
Had the same colors and fonts been used throughout, as in the scatterplot
in Figure 5-1, it'd be a headache to browse this poster.

Sometimes visual hierarchy is used to show process or reflect how you might
explore a dataset. Imagine you generate a lot of charts during the data explo-
ration phase. You make a few graphs to see an overall picture, and in that
summary, you note specifics and then make charts that focus on those. You
can design your graphics to follow this same logic, basically taking readers
on a tour of your analysis.

The bottom line: Graphics that follow a visual hierarchy are easier to read and
can be used to guide readers toward points of interest. In contrast, flat graphics
that lack flow make it harder for readers to interpret results and discourages
closer looks. You don't want that.

READABILITY

An author who uses words to describe a world or character interactions makes
abstractions so that a reader can picture what's going on. Poor descriptions
and little character development challenge readers to make sense of what
seem like obscure clues. If readers can't connect the dots and understand
what the author tries to describe, the words lose their value.

Similarly, you encode data with visual cues when you visualize it, and then
you or others have to decode the shapes and colors to draw insights or to
understand what a visualization represents, as shown in Figure 5-6. If you
don't describe the data clearly, which makes a data graphic readable, then
the shapes and colors lose their value. The connection between the visual
and the underlying data is broken, and you end up with a geometry lesson.