Tips for Effective Data Visualization
February 20, 2019
How can you present data in a new and interesting way? What is data visualization? Is it a better way to present your data? A panel of researchers weigh in on all the new possibilities of data presentation.
Jacob Stuivenvolt Allen
- The first and most important thing to do when presenting: know your audience
- Break the mold
- A figure should stand on its own, and give information that words really can’t. That’s why we use figures.
- Notes on implementing these ideas:
- It’s easy to get bored, distracted, or confused. That’s not the audiences fault.
- Use progressive disclosure (providing only the most immediate, important information for that moment) so people don’t get distracted looking at things you’re not talking about yet
- Include transparent images for things you’ve already talked about on that slide, so you can still see where you’re at and where you’ve come from while staying focused on what’s current
- You control the focus. If you don’t have something on the board, people will focus on you.
- Too many pretty visualizations can be distracting. Focus on one at a time rather than having multiple up there at a time
Alexis Ault
- Here are some key things we tend to forget as we are beginning to compile manuscripts
- Think about how the figures tell the story. First thing you think about in a paper: what are the 3-4 figures that are going to tell this story. What are the five highlights? Then do a brief outline.
- Charts need:
- Labeled Axes
- Data points with distinguishable colors
- A legend
- San Serif Fonts (Arial and Helvetica also work)
- An error bar on “prorogated analytical uncertainties”
- Consistent symbols
- When you transition between figures, the data can be the same but you need a clean, clear legend and different color
- You can visualize the same data differently depending on the key point you’re trying to make
- Multi-type figures
- Build in components but sometimes you need multiple ideas linked together in different figures in a set
- You need letters to signify parts of figures, captions need to be concise and not too specific (if you’re too specific the images lose their clarity)
- Schematic manuscript figure
- Time series, how variables change over time
- Sometimes they don’t contain values, just a picture of what is happening
- When making figures for a manuscript, you want them to be able to stand alone, but there are specific attributes all figures much have (labeled axes, etc.). It takes practice and you have to make multiple drafts and get feedback to get it right. Each iteration will show you what works and what doesn’t work.
Mike Daines
- Usually graphic designers are polishing data, tables and figures to make it beautiful
- Edward Tuft: Father of information design said to use the smallest effective space possible
- Visual Economy: strategic use of space
- Helvetica is the most economic typeface, there’s nothing there to distract you from the message, no visual waste
- We usually write papers in 12 point but read things in 8 point (magazines, textbooks)
- A description with an image looks good and does the job.
- Researchers want things to be BIG and they don’t want anything left out, but you can’t make everything big because it becomes too confusing to read, CHOOSE what you want to have larger
- Be careful in how you use color and using it for emphasis (ex: cool colors w/one warm color)
- Group colors into families (pallets) and use one color that’s different for emphasis
- If it looks boring its probably a good choice for professional data, don’t look for “fun”
- Dont: use lots of competing colors, the same font with no changes for emphasis
Q: Where to start?
- Alexa: Adobe Illustrator, insert a plot from excel and leave it as a layer and creating a new layer (overlay) to recreate it yourself. It takes time.
- Mike: Adobe has really excellent video tutorials