In our visual-heavy society, the way we present data is becoming just as important as the data itself. While technology allows us to create complex visuals easily, many scientists lack formal training in design, leading to figures that are suboptimal or even confusing.
Effective figure design isn’t about following a rigid formula; it’s about making deliberate choices to tell the story of your data. Based on the insights of Stephen R. Midway from Louisiana State University, here are ten sequential principles to elevate your data visualization game.
Phase 1: Figure Design
Before you even touch your computer, focus on the message.
- Principle #1: Diagram First. Prioritize the information you want to share and envision the design mentally or on paper. This prevents software limitations from biasing your visual tools.
- Principle #2: Use the Right Software. Don’t rely on simple spreadsheet programs for complex technical figures. Be prepared to learn new methodologies or software (like R’s ggplot2) to achieve professional results.
Phase 2: Figure Making
This is where you make the technical decisions that define your visual.
- Principle #3: Use Effective Geometry and Show Data. Choose shapes—like bars, dots, or lines—that match your data type (amounts, distributions, or relationships).
Pro Tip: High “data-ink ratios” are best. Always try to include the raw data (like jittered dots) to provide context for the summary geometries.
- Principle #4: Colors Always Mean Something. Color is a powerful tool that should be used strategically.
- Use Sequential schemes for increasing values.
- Use Diverging schemes for two extremes.
- Use Qualitative schemes for group differences.
- Ensure your figures work for colorblind readers and remain effective when converted to grayscale.
- Principle #5: Include Uncertainty. Failing to show uncertainty can be misleading. Use error bars, shaded intervals, or distributional geometries like box plots to represent the reliability of your data.
- Principle #6: Panel/Small Multiples. Repeat a figure across a grid (faceting) to highlight differences across variables like time or groups. This makes the data inevitably comparable.
- Principle #7: Distinguish Models from Data. Raw data and inferential statistics (like a fitted regression line) are different messages. Ensure the reader can distinguish between the two and backtrack the model details.
- Principle #8: Simple Visuals, Detailed Captions. Captions should be “standalone,” meaning the major points can be understood even if separated from the main text. Explain every geometry and symbol used.
Phase 3: Figure Review
Final steps to ensure your message hits the mark.
- Principle #9: Consider an Infographic. If you need to reach a broader audience or increase memorability, blending text and diagrammatic elements can be more effective than a standard figure.
- Principle #10: Get an Opinion. Seek external reviews from colleagues. Have them review the figures without the text to see if the visual message is truly clear on its own.
Why It Matters
A figure maker has only a few seconds to make their point. The longer a reader struggles to understand a visual, the more likely they are to lose interest or misinterpret your work. By following these principles, you ensure your scientific message is disseminated with clarity and impact.