Humans are highly visual creatures. From traffic lights to weather maps, we are surrounded by messages that tap into our ability to quickly absorb and interpret visual signals. Science is no exception. Data visualizations, such as graphs, help scientists view multiple pieces of data simultaneously to decipher patterns and trends. For example, below is a visualization that reflects the number of respiratory illness cases reported in different US states; beside it is an excerpt from a data table that contains some of the same information. You likely find that the map is easier to make sense of than the table, with your eye naturally distinguishing which parts of the country reported elevated levels of illness due to their darker, redder hues. Additionally, the map takes up much less space than the full table, which is 56 rows long.
A map showing the levels of respiratory illness reported in each US state. In this visualization, color is used to reflect the number of cases reported in each state over the course of a week. Below it is an excerpt of a table (the full table is 56 rows long) that contains the same information. (source)
Visual communication is also a key tool when scientists share knowledge with a general audience, such as guidance on how to prevent spread of a disease. Communicating visually, rather than relying solely on text, can often make a message accessible to a larger audience by circumventing language barriers. In addition to making information more understandable, visual communication can also draw on a rich lexicon of cues to achieve specific effects, e.g. directing a viewer’s attention to a notable datapoint or evoking a particular emotion through deliberate use of color.
Illustrated guidelines for how to wear a mask effectively to reduce spread of SARS-CoV-2. In addition to showing the correct way to wear a mask, the image uses colors and iconography (green checkmark vs. red “no” symbol) to distinguish correct vs. incorrect usage, and the blue vs. gray backgrounds further reinforce the distinction. This image communicates its message without reliance on text. (source)
There are several different visual approaches that can be found in science communication. Data visualizations, mentioned above, summarize data in a visual form (e.g., through graphs) to make it easier to analyze and understand. Photos and videos provide a real-world view of a subject or situation. Illustrations depict structures, concepts, and processes — often in a way that is impossible to capture in a photograph — and highlight relevant details. Infographics can combine several of the visual approaches mentioned above along with narrative text, often with the goal of giving a simplified overview of a broad subject.
Examples of different visual communication approaches used in science and public health. This data visualization allows the audience to quickly determine that tuberculosis cases are decreasing over time, while the photograph provides a real-world example of the severity of the disease. The illustration intuitively depicts how the disease is spread, while the infographic provides several pieces of information in an easy-to-digest format. (source)
It is increasingly common for visual communication to also include interactive elements, which enable a user to choose which pieces of information they wish to view, and how. An example of an interactive visualization is the Johns Hopkins COVID-19 Dashboard, active from 2020 to 2023, which allowed web users to zoom in on specific areas of a world map to view the number of COVID-19 cases in individual cities.
With the multitude of visual approaches available, we should keep in mind that differences in vision mean that not everyone is able to see a visual message the same way. Accessibility is a necessary consideration and can be supported through a variety of approaches, which include colorblind-friendly color palettes and informative captions that can be processed by screenreaders.
The wealth of resources and capabilities that make visual communication a useful tool also make it a double-edged sword. It can be easy to selectively display data to advance a particular agenda, or to make a dishonest message appear legitimate by embellishing it with sleek imagery. The growing power of AI image generation means that even photos and videos can be convincingly falsified. As a result, we need to carefully examine the information presented in a visual, especially during an outbreak or during other challenging events when accurate information is crucial.
Here is an example of a visualization that contains important information — the magnitude of infectious disease outbreaks over the course of human history — but can give an inaccurate impression of the data. The scale shrinks as the timeline goes further back (imitating the perspective that one might see in a photograph or drawing) to evoke a sense of distance. However, this element makes it difficult to compare the magnitudes of outbreaks at different points in time because outbreaks in the distant past appear smaller than those in recent times. For example, the sizes of circles suggest that the Bubonic Plague caused fewer deaths than the Spanish Flu, but the numbers reveal that this is not the case. Even visualizations made by respected sources with good intentions need to be examined critically before drawing conclusions from them.
A history of infectious disease outbreaks.
The size of circles is related to the number of deaths in each outbreak, but the use of perspective makes outbreaks further in the past appear smaller (source). Notably, the original version of this image included an additional panel below that compares different outbreaks with the correct scale.
So, what exactly can we look for when examining a visual to ensure that we’re not being misled? Some questions that you can ask yourself include:
Is the visualization presenting actual data, or a schematic of what data might look like?
What is represented by the space occupied by the visualization, e.g. where do the axes start and end, what do they represent, and what is their scale?
How is uncertainty represented, if at all?
What do other parameters in the visualization (colors, sizes of data points) represent? Are they redundant with other elements? Do they tell a different story?