Storytelling with Data
The importance of context
- Exploratory vs explanatory graphics.
- Exploratory is for the analyst.
- Exploratory is for the audience.
- Who is the audience, who is the stakeholder?
- What do you need your audience to know or do?
- How will you communicate to your audience? Live or written by email?
- What tone do you want your communication to set?
- What data is available that will help make your point?
- If you had 3 minutes to tell a story, what would be the big idea : the unique point of view in one sentence?
- Let’s now move to storyboarding.
Choosing an effective visual
- Go with a simple text or a figure.
- Available charts: scatterplot, table, lines, slopegraph, vertical bars, horizontal bars, stacked vertical bars, stacked horizontal bars, waterfall, square areas.
- Table? Never in a live presentation (too dense).
- Heatmap? Never! A better alternative is a slopegraph.
- Color saturation is for visual contrast: put everything in B&W except the one thing you want to draw attention on.
- Speak in relative, not in absolute: scatterplot, average, percentage, waterfall (change, increment), areas for comparing categories.
- Avoid clutter and 3D, avoid saturated series and secondary axes.
- Prefer a zero baseline and equal scales among several graphics.
- Prefer square forms over round forms.
- Avoid pie charts or donut charts.
- Prefer a horizontal bar chart.
Clutter is your enemy !
- Avoid cognitive load: a simple chart, a few line, one image.
- Avoid borders (limited to the L axes), avoid gridlines and background shading.
- The y-axis is often facultative.
- Clean up axis labels; summarize labels (Jan vs January)
- Bold and colors is for making the important stand out.
- Limit to 2 or 3 colors.
- Add white space.
- Add details to the important areas (annotate).
- Add the legend on the graph, not outside.
Focus your audience’s attention
- Use preattentive attributes in text : bold, italics, separate spatially, underlined, more tone or color (B&W), data stand out, numbered data, size.
- Avoid shades of red and green.
- Use preattentive attributes in visual : difference in orientation, shape, length, width, size, curvature, added marks, enclosure, hue, intensity, spatial position, motion
- Position on the page follows the 1-2-3-4 loop.
- Consult:
- Cultural Color Connotations, David McCandless.
- The Visual Miscellaneum : A Colourful Guide to the World’s Most Consequential Trivia.
Think like a designer
- Eliminate distractions.
- Not all data are equally important.
- When detail is not needed, summarize.
- Would eliminating ‘this’ change anything?
- Push necessary, but non-message-impacting items to the background.
- The power of subcategories, subdivisions.
- Legible, clean, straightforward language.
- Action titles.
Lessons in storytelling
- Write a story with a schema (in 3 acts).
- Write a narrative structure (the hero’s journey).
- Use the power of repetition: 3, 7, 12, 30, 40, 100, 500, 1000.
Final thoughts
- Learn your tools well.
- Google Spreadsheet, Tableau, R, D3 (JavaScript), Processing, Python, etc.
- Iterate.
- Devote time to storytelling with data.
- Seek inspiration throught good examples.
- Find your style.
- Consult websites.
- Eagereyes, Robert Kosara.
- FiveThirdyEight’s Data Lab.
- Flowing Data, Nathan Yau.
- The Functional Art, Alberto Cairo.
- The Gardian Data Blog.
- HelpMeViz, Jon Schwabish.
- Junk Carts, Kaiser Fung.
- Make a Powerful Point, Gavin McMahon.
- Perceptual Edge, Stephen Few.
- Visualising Data, Andy Kirk.
- VizWiz, Andy Kriebel.
- Storytelling with data, Cole Nussbaumer Knaflic.
- and many more.
Wrap up
- Understand the context.
- Choose an appropriate visual display.
- Eliminate clutter.
- Focus attention where you want it.
- Think like a designer.
- Tell a story.