[GUEST POST] Don't Know Where to Begin with Data Design? Easy Strategies to Get You Started.

Echo's Introduction: There is a large, and growing, community of science communicators. The purpose of my website is to provide resources to help other researchers, evaluators, academics, and scientists effectively share their work. So, I recently decided to make my blog available for guest posts. This is a win-win: Giving others an opportunity to do #scicomm, while also bringing new voices and perspectives to my audience. If you're interested in writing a guest post, email me at echo@echorivera.com.


[GUEST POST] Don't Know Where to Begin with Data Design? Easy Strategies to Get You Started.

Written by Maureen Berg, @MaureenBug

Data visualization is an important part in communicating science, but it can be overwhelming if you’re starting from scratch. Because you know your data so well, it is easy to overlook small, clarifying details that would help your audience understand your results. I wrote this post to help others improve their data visualization strategies by focusing on the the commonly overlooked details.

The Do's and Don'ts of Bar Graphs

dataviz_template-barplot_DO.png

Make fonts easy to read and consistent - font sizes should be large enough to easily read, and should be used consistently. Helvetica, Courier, Arial, and Verdana are good examples, and are some of the easier fonts for people who are dyslexic to read.

Properly show your data (e.g. include error bars) - label axes and include units. When appropriate, show error bars or confidence intervals, and denote statistical significances.

Keep the design clear and simple - overall, your dataviz should be accessible to your readers, so they don’t need to spend a lot of time and energy in understanding it. Using a simple color scheme, and eliminating distracting elements will help your reader quickly understand the important data trends.

(Don't) Use fonts that are hard to read and are misaligned - Data text should not be overlapping with other parts of the plot, and vertical text should always be avoided (use a different chart type or small multiples if needed). Align all your text so your figures look clean and professional, and so misaligned text isn’t distracting to your audience.

(Don't) Mix and match fonts - Use one consistent font throughout all your data. In some cases, using two fonts is beneficial (e.g., to establish a design hierarchy); if two fonts are being used, keep their use consistent.

(Don't)Use color gradients or drop shadows - avoid using color gradients for categorical values (e.g. species of trees in a forest), drop shadows, and unnecessary borders. Colors gradients should only be used with continuous values, like you’d see in heatmaps (see below for examples). Basically, try to avoid elements that are not clean or “flat” design. These are visually distracting because they are not important for communicating your data.


The Do's and Don'ts of Scatter Plots

Make your color and shape legend scheme simple - with two categories, using both different colors and shapes are not necessary, and add visual clutter. Only use both colors and shapes when you’re working with multiple, intersecting categories (e.g. weights of cats vs dogs in USA, Canada, and Brazil). Even then, other design strategies might be more effective, like using colors (e.g., a color + gray) to emphasize key points.

Make your data points visible - the point size should be big enough to comfortably see if your readers are skimming through your data; they shouldn’t need to strain their eyes.

Choose an appropriate axis scale - your data should more or less fill out the plot area so any trends in your data can be easily seen.

(Don't) Use a complicated shape and color scheme - too many legend labels can be confusing for readers if they are not necessary. Data that only has two categories can be shown using either colors or shapes.

(Don't) Use colors and sizes that are hard to see - data points that are too small and are neon/light colors cause eye strain.

(Don't) Make your axis scale too large or too small - if your scale is too small, you’ll cut off some of your data. If your scale is too large, some of your data trends might not be visible, and the extra whitespace can be distracting.


The Do's and Don'ts of Heat Maps

For continuous data values, use a color gradient with a scale that will capture differences within your data - using a good color gradient is an excellent way to visually show the trends. Choose a wide enough scale so small differences in your data values can be seen with an appropriate color change.

Choose colors that are distinguishable - choosing colors that are not similar, such as blue and orange, will allow readers to easily understand your data trends.

(Don't) Use colors that cannot be distinguished by people who are colorblind- avoid color pairs such as green and red. If using Adobe, you can turn on a color blind filter to see if your colors are distinguishable; go to View > Proof Setup > color blindness. Good palettes for color blind people can be found at Color Universal Design and Color Brewer. 

(Don't) Use a rainbow gradient - Rainbow gradients have too many colors that, while they may show small differences in your data, they are confusing and distracting for readers. It takes too much time for readers to study and understand the color scale before they are able to interpret the data. Colors should make important data trends “pop out” for readers.

Wrap-up

Good data visualization helps your audience easily understand your results. Once you’ve made the first drafts of your data figures, it is helpful to go back over them with these tips in mind. I find it helpful to see what clarifications a friend or colleague needs to understand your data without you verbally explaining it to them. I then try to alter the design to improve the clarity of my results. A little effort to improve your data visualization can go a long way.

Grab the Downloadable Infographic of This Post! >>


About Maureen:

Maureen is originally from Cincinnati, Ohio, and received a BS in Biology at the University of Dayton. Maureen recently graduated from UC Berkeley with a PhD in Integrative Biology, and is currently applying for and interviewing for positions in the Bay Area. She has always had an eye for design, and has carried over that passion into her research. During graduate school, Maureen knew many people who didn’t prioritize good data visualization to the extent she did (especially in informal setting, such as lab meetings), so she decided to put together a simple guide to show how easy it is improve the clarity of your results by changing a few visual aspects of your figures. You can follow her on twitter @MaureenBug.