8 Quick Tips for Better Plots
If the “story” is simple, keep it simple. If the “story” is complex, make it look simple. Tell the truth – don’t distort the data.
8.1 Get Rid of Chartjunk
An easy way to instantly improve your plots is to eliminate superfluous material. Extra tick marks and grid lines; unnecessary text and arrows; decimal places beyond the measurement error of the level of difference; cute little butterfly clipart: this chartjunk has no meaning and it clutters up a chart, making it hard for your viewer to see what’s most important - your data. The amount of ‘ink’ on your figure should directly correspond with the amount of data you present. If it doesn’t, you have a lot of chartjunk. (Evidence #10298 that pie charts are never a good choice.)
Try these de-cluttering steps to improve your charts:
- Shift from center to left-justified text
- Retain whitespace
- Clear contrasts
- Remove chart borders
- Remove (or strongly mute) gridlines
- Remove data markers and point labels (unless they are important)
- Remove unnecessary polygon filling
- Cleanup and rename axis labels to be intuitive
- Replace the title with something informative
- Label the data directly using the principle of proximity
- Leverage consistent color and other aesthetics
Creating visual order and reducing chartjunk will dramatically improve your graphic by helping your data stand out.
8.2 Facilitate Comparisons
- Avoid having the graph elements interfere with the data
- Juxtapose or supepose plots (using the same scales)
- Use visulally prominent symbols
- Avoid over-plotting; try jittering, or smooththing
- Don’t change a scale mid-axis
- Use only one scale on one axis
- Use color, judiciously
- Avoid jiggling the baseline
- Don’t distort the data; take care when selecting the encodings
A common mistake is to use more encodings than there are dimensions of the data. If you data only has two dimensions (say number of students in STEM by gender identity), your figure could reasonably use points, rarely area, and never volume. (I’m looking at you, 3D pie chart.)
8.3 Create Information-Rich Plots
Data visualizations cannot exist without text. They require context to infer meaning. Ask yourself:
- Does the caption describe what has been graphed? Does it draw attention to the imporant features? Describe the conclusions drawn by the graph?
- Are the legends and labels clear and intuitive?
- Are important reference lines and points labeled?
Just as an author edits before publishing the novel, and an artist sketches before making the masterpiece, plotting is an iterative process. Proofread for clarity and consistency. Check whether your plots pass the expressiveness and effectiveness tests. Does a viewer draw the same conclusions from the figure that you do?
Here’s a [cheatsheet checklist from this reader](https://docs.google.com/document/d/1rwoa8XPgvTKHE5Weh9RhrO8igfJ6E8bqyZtddZpwomc/edit?usp=sharing] to help you design and improve your data visualizations. Happy plotting!