Color Me Intrigued: Communicating Results Efficiently and Effectively With Color

Pie charts? Three-dimensional graphics? Misleading axis choices? We often hear all about what not to do when visualizing our data. But what makes a good visualization good? In his classic volume, The Visual Display of Quantitative Information, statistician Edward R. Tufte quips that graphical excellence means “that which gives to the viewer the greatest number of ideas in the shortest time with the least ink in the smallest space.”

As laboratorians, communicating our findings is an integral part of our jobs. Doing so effectively means getting the most out of the budget of time and attention your reader has allocated to you. In this article, we will provide a brief overview of the tips and tricks behind some key principles in graphic design applied to laboratory medicine, with the goal of equipping you with the tools you need to create more effective figures and visualizations.

The Science of Design

In its essence, effective visualization catalyzes the process of turning data into information — and information into knowledge. To achieve this, we first must take a foray into the science of perception to uncover how we encode visual information. These concepts extend beyond the basic needs of adequate contrast between colored figures, or fonts large enough for deciphering from a distance. Though striving for visually pleasing color palettes and formats are also desirable elements of a presentation, a pleasing display does not always yield the most efficient results. In other words, there is a “right” and “wrong” way to display data.

Representing Numbers as Colors With Color Maps

Color as a representation of measurement is one of the most common ways to represent a measurement. When using color, besides the oft-overlooked element of ensuring that those with color vision deficiencies (CVD) can still glean insights from our visualizations, it is also our responsibility to convey data in ways that are both efficient and free of manipulation. This requires careful consideration of how the numerical inputs in our data are being represented as colors through a color map. Figure 1 highlights just how important the selection of a color map can be (1).

Unfortunately, the most recognizable color map, called rainbow (or jet, as in the figure), distorts the underlying image because of its sharp demarcations at the transitions between blues, yellows, and reds. More scientifically, the rainbow color map is not perceptually uniform: The variation in shade and lightness is not weighted equally in our eyes, leading to distortions in the representation of our data (1).

Beyond the aesthetic aspects of proper color map selection, there are also efficiency and effectiveness considerations. Borkin and colleagues have demonstrated the danger of the rainbow in their work evaluating color maps used for identifying vulnerable regions on angiogram visualizations in patients with coronary artery disease (2). Their results showed more errors and longer read times for the rainbow color map when compared to a more appropriate, diverging color map (2).

Fortunately for us, the same authors that highlighted the shortcomings of the rainbow color map in Figure 1 have provided a solution. Fabio Crameri’s scientifically derived color maps (www.fabiocrameri.ch/colourmaps), which are known as batlow, present data in a perceptually uniform, CVD-accessible, reproducible, and even citable manner (3). Crameri also has made it remarkably simple to start using these color maps in your own work with a straightforward user guide (www.fabiocrameri.ch/colourmaps-userguide) and masterclasses through Undertone Design (www.undertone.design).

Practical Advice to Improve Your Visualizations

So, with these principles now in mind, what practical advice can we offer now that you are ready to display your data? We propose a series of four steps to get the most out of your figure-making:

1. Think about the story you’re trying to tell.
2. Plot the data in gray.
3. Add color intentionally to highlight key story elements.
4. Solicit feedback often.

Think About the Story of Your Data

First, think about the story you need your figure to convey. Are you trying to explain a key finding, or help readers explore the data on their own?

Explanatory figures should provide only the data that is necessary and sufficient to support a conclusion, such as plotting medians or distributions, while exploratory figures ought to present as much of the data as feasible for the given medium.

Explanatory figures often are best for time-bound, goal-oriented presentations to multiple audience members in person, such as lectures, quality improvement meetings, and proposals. In contrast, when readers can devote as much — or as little — time as they would like to extract information from a figure, an exploratory figure can be much more powerful. Appropriate settings may include research articles, quality assurance reports, and dashboards.

Start With Gray, Then Add Color

Next, besides the key points to be gleaned from the data, what is it that you want your audience to notice first? Usually, there is an order in which it makes the most sense to understand a figure. We can augment that perception by using the idea of starting with gray. Figure 2 represents a hypothetical healthcare system with four hospitals, A–D, each with their own laboratories. You have been tasked with presenting data about testing volumes to a set of hospital stakeholders and business leaders. Your color choices can have a clear impact on the first, and lasting, impression that your audience takes from the data being shown.

By first plotting all the data in gray, you must make intentional color choices to highlight key points. Figure 2 shows two possible options that depend entirely on the story you aim to tell. Bottom left displays the more exploratory color choices, where each hospital is represented as its own discrete color within the batlow color map. This figure underscores the point that testing volumes across all four hospitals are increasing.

However, if you instead wanted to highlight hospital C’s outlier status, perhaps in a pitch to fund more technologist positions, you could keep the other hospitals gray, while highlighting only hospital C in color (bottom right).

Solicit Feedback

Finally, solicit feedback often: can viewers interpret your data quickly, without extra prompting or explanation from you? Do they recognize the most important points you are attempting to highlight? If not, adjust and repeat. Making intentional design and color choices can help to get the most out of your readers’ attention, resulting in quick and accurate data interpretation.

Following the strategies outlined in this article can help you streamline the process from raw data to actionable insights, ensuring that your visualizations not only capture attention but also convey your message with clarity and precision. With the right tools and approach, you can elevate the impact your visualizations have on your audience, and we look forward to seeing how they enhance your next project.

Carly Maucione, MD, is a resident in clinical pathology at the Washington University School of Medicine in Saint Louis, Missouri. +Email: [email protected]

Nicholas C. Spies, MD, is the chief resident in clinical pathology at the Washington University School of Medicine in St. Louis, Missouri. +Email: [email protected]

References

1. Crameri F, Shephard GE, and Heron PJ. The misuse of colour in science communication. Nat Commun 2020; doi: 10.1038/s41467-020-19160-7.
2. Borkin MA, Gajos KZ, Peters A, et al. Evaluation of artery visualizations for heart disease diagnosis. IEEE Trans Vis Comput Graph 2011; doi: 10.1109/TVCG.2011.192.
3. Crameri F. Scientific colour maps 8.0. https://www.fabiocrameri.ch/colourmaps (Accessed November 2023).