66  Explore versus Explain

The first important aspect of visualization is to recognize the distinction between exploratory and explanatory data visualizations, summarized in ?fig-Purpose and Table 66.1.

Exploratory visualizations
are among the first steps of any Exploratory Data Analysis (EDA) workflow. Here, data visualization is not only an extension of descriptive statistics, but also use for diagnostic plots. It allows us to get the first, crucial impression of a data-set, and to assess if we’ve used the appropriate statistics.
Explanatory visualizations
are the polished plots that appear in scientific writing and presentations. Data journalism is an extension of this, where the audience is typically lay people, and further still infographic are another elaboration. Here, I’ll focus on explanatory visualizations by and for scientists.

Figure 66.1: The role of Data visualization in Exploring and Explaining Data. The scientist has the job of using visualizations to not only explore their own data, but also explain it to a specific audience. Figure adapted from an article by Bang Wong
Exploratory Explanatory
When First stages of analysis End of analysis
Purpose Exploratory Data Analysis (EDA) and diagnostic plots Communicate a clear message to a specific audience
Effort Quick & dirty Labor intensive
Data content Potentially many variables Edited to meaningful variables
Audience Small, specialist only Broad, specialists to generalists
Breadth Internal, for yourself & colleagues External, for publication & presentation
Table 66.1: The two broad purposes of data visualization.

66.1 Overarching themes

Continuing with an overarching theme throughout this book, we’ll see how these two sides of data visualization mirror how we use writing, sketching, discussions, presentations, and other forms of communication as a tool to not only transfer information, but to also help us to understand that information ourselves.

We’ll focus on explanatory data visualizations. I view these as an extension and refinement of exploratory data visualizations. I’ll mostly limit our discussion to static print and screen plots, which make up the lion’s share of scientific data visualizations. Examples of interactive plots will be presented in XX. For tips on presenting quantitative data, e.g. in an oral presentation, see XX.