Appendix F — Chart production tips

Building charts is a part of the data exploration process, and often the only audience is ourselves or our editors. We’re just trying to learn something about our data. At title might help us quickly understand what we plotted, but we might not sweat every axis name and every detail.

But we can also build charts for others, like our readers or an audience that hasn’t been mired in the data. It may run with a story or be part of a report. With these you are specifically trying to communicate a finding from your data in far more detail to someone who has far less context to begin with. Whether you are building this in ggplot, Datawrapper or some other tool, you must take more care to be thorough, accurate and communicative.

F.1 Titles, descriptions and annotations

Chart titles and descriptions can be some of the most difficult writing you can do as a journalist. You don’t want to describe the steps of your analysis nor say the obvious, but you do need to give the reader all the relevant detail needed to understand the chart. Write titles, descriptions an annotations as if the chart stands alone, and a reader knows nothing before viewing it.

Some tips paraphrased from Nathan Yau’s Data Points book …

• The title – typically larger and bolder fonts — sets the stage or describes what people should see or look for in the data. A descriptive title also helps. For example, “Rising Gas Prices” says more about the chart than just “Gas Prices.” By including “Rising”, it presents the conclusion immediately, and readers will look to the chart to verify and see the details. Saying just “Gas Prices” leaves the data interpretation to the readers and places them in the exploration phase.
• The description or lead-in text is used to prepare readers for what a chart shows, but in further detail. The text expands on what the title declares, where the data is from, how it was derived, or what it means (best charts do this, says @crit). Basically, it’s information that might help others understand the data better but often doesn’t directly point to the specific elements.
• To explain specifics points or areas, you can use lines and arrows as an annotation layer on top of a chart. This places descriptions directly in the context of the data so that a readers doesn’t have to look outside a graph for additional information to fully understand what you show.

F.2 Other considerations

• Proper data encodings and visual cues: Think about what you are trying to convey with the graphic and plot your data in a fashion that furthers that understanding. (See Nathan Yau’s Data Points, Chap. 3.)
• Legends for encodings: If your plots include labels for categories on the chart, you may not need a separate legend, but be sure readers can distinguish items.
• Labels for axis: They help describe the value being plotted. In some obvious cases where the meaning in clear, like years, they may be dropped.
• Include unit values to further understanding: Sometimes the value can be added to the plot itself, other times grid lines may be enough.
• Annotations: Add explanations to the plot if they help readers understand nuance of what you area trying to convey.
• Source of the data: This is the course of the data, not the delivery method. (i.e., The Comptroller of Public accounts, not data.texas.gov.)
• Your byline: Credit yourself and your publications.