Adding Context

11.5. Adding Context

When preparing a visualization for sharing more broadly, it’s important to add additional context to the plot. We can do this with reference markers, color, and labels. We often aim to give enough context in our plots so that they can stand alone—a reader should be able to get the gist of each plot without needing additional explanation elsewhere. That said, every element of a statistical graph should have a purpose. Superfluous text or plot features, often referred to as “chartjunk”, should be eliminated. In this section, we provide a brief overview of ways we can add helpful context to our plots and an example where we create a publication-ready plot by adding context.

  • Reference Markers. Reference points and lines can provide benchmarks, historical values, and other external information to compare our data against and help interpret the results. For example, we often add a reference line with slope 1 to a quantile-quantile plot. We might also add a vertical line on a time-series plot to mark a special event, like a natural disaster.

  • Labels. It is good practice to consistently use informative labels on tick marks and axes. For example, axis labels often benefit from including units of measurement. Our graphs should contain titles and legends when needed. Informative labels are especially important for plots that other people will see and interpret. However, even when we’re doing exploratory data analysis we often want to include enough context so that when we return to an analysis we can easily figure out what we plotted.

  • Captions. Plot captions serve several purposes. They describe what has been plotted and orient the reader. Captions also point out the important features of the plot and their implications. It’s okay for the caption to repeat information found in the text. Readers often skim a publication and focus on section headings and visualizations so plot captions should be self-contained.

The following example demonstrates how to add these context elements to a plot to make a publication-ready plot.

11.5.1. Example: 100m Sprint Times

The figure below shows the race times in the men’s 100-meter sprint since 1968. These data include only races that were electronically timed and held outdoors in normal wind conditions, and the times included are only for those runners’ that came in under 10 seconds. The plot is a basic scatter plot showing race time against year. Beginning with this plot, we augment it to create a plot featured in a 538 article about the Olympic 100-meter sprint [Planos, 2021].

fig = px.scatter(run, x='year', y='time', hover_name='name',
                 width=450, height=250)

When we want to prepare a plot for other people, it’s important to first think about the plot’s takeaway message. In this case, the main story is that Usain Bolt’s world record time of 9.58 seconds set in 2009 is still untouched. In fact, the second-best race time also belongs to Bolt. We provide context to this plot by adding a title that directly states the main takeaway, units of measurement in the y-axis label, annotations on key points in the scatter plot, including the two best race times that belong to Usain Bolt, and a caption describing the data. In addition, we add a horizontal reference line at 10 seconds to clarify that only times below 10 seconds are plotted, and we use a special symbol for the world record time to draw the reader’s attention to this crucial point.

title = to_plotly_text('''
No one is coming close to Usain Bolt's best times
<sup>All times under 10 seconds in the outdoor men's 100-meter sprint,
included are times from only electronic readings and under regular wind conditions</sup>

fig = px.scatter(run, x='year', y='time', hover_name='name',
                 width=550, height=450)
fig.update_yaxes(range=[9.5, 10.05], title="time (seconds)")
fig.update_xaxes(range=[1966, 2026])
margin(fig, t=70)
fig.update_layout(title=dict(text=title, y=0.95, x=0.1))

fig.add_hline(y=10.0, line_width=2, line_dash="dash", line_color="black")

hines = to_plotly_text('''
At the 1968 Mexico City
Games, American
Jim Hines was
the first sprinter to break
the 10-second barrier


fig.add_annotation(x=1968, y=9.95,

fig.add_annotation(x=2009, y=9.58,
                   text='Usain Bolt set the world<br>record in 2009...',

fig.add_annotation(x=2012, y=9.63,
                   text='...and the<br>Olympic<br>record in 2012',


These bits of context describe what we have plotted, help readers see the main takeaway, and point out several interesting features in the data. This plot can now be a useful part of a slideshow, technical report, or social media post. In our experience, people who look at our data analyses remember our plots, not paragraphs of text or equations. It’s important to go the extra mile and add context to the plots we prepare for others.

In the next section, we move on to specifics on how to create plots using the plotly Python package.