When Graphic Design meets Science: Displays of Evidence for Making Decisions

Exploring the connections between entirely different fields has been one of the most intriguing and fulfilling aspects of my university journey. The intersection of graphic design and science, particularly when it comes to representing data in a truthful, credible, and precise manner, has continually fascinated me. In this blog post, I’d like to delve into the foundational principles of effective data visualization, with a specific focus on the 2nd chapter of Edward Tufte’s book, Visual Explanations: Images and Quantities, Evidence and Narrative, his analysis of the case from 1854 and my analysis of the research data representation today.

As I navigate through my current computational social science class, I find myself reading several scientific papers each week. In this field, the art of data representation and visualization can indeed make or break a paper’s chances of being published in prestigious scientific journals. Edward Tufte, a prominent voice in the realm of data visualization, passionately emphasizes the substantial impact that the quality of methods employed in presenting and assessing quantitative data can have on a study’s overall outcomes. In other words, a well-crafted visual representation has the potential to ignite action and shape the course of scientific research.

The immediate link between the data representation as both design and scientific problem appeared to me when I realized that I’ve heard a “John Snow” joke twice in one week – in my CS elective class and in Graphic Design class. In 1854, Snow successfully determined the causal link between water and cholera outbreak, which back then was considered to be caused by miasma. Snow drew a map (Fig. 1), where he demonstrated the location of London’s 13 community pump-wells and cholera deaths by stacked bars perpendicular in their location. His graphics helped to contribute in removing the pump-handle, that helped to avoid new outbreaks in the future. With this example, Tufte suggests four points that highlight why Snow’s method was good. I will be discussing some of them as I am analyzing the map from a research paper in 2023.

Today, research has evolved significantly since the year 1854. Over decades of academic development and rapid technological advancement, the landscape of research has undergone a big transformation. One of the differences is the way researchers approach causation. Nowadays, researchers exhibit a higher level of caution when making causal claims due to the big number of potential confounders that need to be considered.

One, in my opinion, good example of the power of data visualization in contemporary computational social science papers comes from the research on global digital inequality conducted by NYUAD faculty members Moumena Chaqfeh, Rohail Asim, Bedoor AlShebli, Talal Rahwan, and Yasir Zaki. Figure 2, a visual representation in their research, effectively summarizes the outcomes of their experiment. In this graphical depiction, circles denote various locations, colors convey average page load times, and diameters represent adjusted costs per Gigabyte. This visualization illustrates a stark global digital divide that persists today.

In contrast to the conventional approach of presenting research findings through a myriad of separate graphs, regressions, and charts scattered throughout the text, the authors of this study have adopted a more effective approach. They have consolidated these diverse elements onto a single map, presenting all the necessary information cohesively. This approach successfully reveals a correlation between Internet quality, pricing, and geographical location in the developing world, even though it does not explicitly establish a causal link. This presentation not only facilitates quantitative comparisons but as authors then propose possible solutions, due to the good graphics this paper received a considerate press coverage, which increases chances of the findings having an actual impact.

Moreover, the current paper effectively implements a concept highlighted by Tufte, focusing on the “assessment of possible errors in the numbers reported in the graphic.” Similar to Snow’s approach, these potential errors are systematically addressed through detailed comments and, in contemporary research, by acknowledging and addressing the study’s limitations.

A short note about Causal Diagrams:

In “The Book of Why,” written by Judea Pearl and Dana Mackenzie, authors did a new visualization of John Snow’s pioneering work in investigating the cholera outbreak using causal diagrams. (Figures 3 and 4) These diagrams provide a structured representation of the confounders and causal relationships at play, offering a comprehensive view of the complex interplay of factors.

As emphasized by Tufte, “descriptive narration is not a causal explanation,” and this holds particularly true when we attempt to explain intricate causal inferences. Conventional X-over-time graphs often fall short in effectively conveying these nuanced relationships. This is where the beauty of causal diagrams shines through. They offer a direct and intuitive representation of the cause-and-effect relationships at the heart of a phenomenon, demonstrating causal inference quite intuitively.

Conclusion:

As I am currently working on a computational social science paper this semester, Edward Tufte’s chapter has offered me a fresh perspective on the graphics our team is incorporating into our research. Tufte’s approach to data visualization emphasizes analysis and presentations that aim to reveal truth, rather than serving a particular viewpoint or agenda. He assumes that the audience is willing to actively engage with the content and invest the necessary effort to understand it. With a focus on employing effective methods and drawing inspiration from successful examples, I am excited about the prospect of enhancing the graphic design aspects of my scientific work.

‍

References:

Tufte, Edward. Visual Explanations (Graphics Press 1997) – 2nd Chapter

The Book of Why: The New Science of Cause and Effect, by Judea Pearl and Dana Mackenzie (2018). Basic Books. ISBN: 978-0465097609

Chaqfeh, M., Asim, R., AlShebli, B., Zaffar, M. F., Rahwan, T., & Zaki, Y. (2023). Towards a World Wide Web without digital inequality. In Proceedings of the National Academy of Sciences (Vol. 120, Issue 3). Proceedings of the National Academy of Sciences. https://doi.org/10.1073/pnas.2212649120

‍