How do you balance simplicity and complexity in your data visualization?

When visualizing your data, it's often helpful to think of the key take home messages for each visualization. Then, think about how you might get the audience there. This works well for both simple and complex visualizations.

A good data visualization isn't a monologue—it's a conversation. A conversation that transcends numerical figures and enters the realm of insight and understanding. But often, we get so engrossed in the "how" of data visualization that we overlook the "why." Is it enough to know your audience's current needs, or should we take it a step further? Once you understand your audience, try anticipating what they might need to know next.

Your purpose defines the message you want to convey, whether it's to inform, persuade, or inspire action. Understanding your audience helps tailor your visualization to their needs, ensuring it resonates with them. In my approach, I always begin by asking, "What is the main takeaway I want my audience to have?" This clarity of purpose guides every aspect of the visualization, from the choice of chart type to the selection of data points. Furthermore, understanding the audience's background, knowledge level, and preferences is crucial. Are they experts in the field, or are they new to the topic? Do they prefer concise visuals or are they comfortable with more complexity? These insights shape the design and complexity of the visualization.

A great considering for data visualisation, in addition to the tips by others below, is to consider data like a film director. You have the script, you have the vision. You know the beginning and ending, and you know the market. Therefore, consider that your data is like the script. You have to frame it in such a way that the audience will be able to see what you see, for the tone you want to take. In other words, start by considering the elements of what the data represents, what elements are useful, and what aren't. Cut what isn't needed, and present only what matters to ensure your vision matches the audience.

Balance simplicity and complexity in data visualization by tailoring to audience understanding, highlighting key insights without clutter, using visuals judiciously, aggregating data meaningfully, guiding with a clear narrative, gathering feedback for refinement, and iterating for continual improvement.

Choosing the right type of chart is key to balance simplicity and complexity in your data visualization for which it is crucial to understand the scale of measurement used in the data (nominal, ordinal, interval, or ratio) as this will influence the chart choice. For instance, histograms are ideal for interval or ratio data, while nominal or ordinal data is better represented by bar charts. It's also important to be wary of chart types that can distort perception, such as 3D charts or pie charts with too many slices, which can mislead the audience. Ultimately, the goal is to enhance the data's narrative by making it accessible and understandable, without oversimplifying or overcomplicating the visual message.

An old sociologist once told me, as a joke: if you have to explain the chart, you've likely already lost the plot, and you're leaving with pie on your face. Choosing the right chart can be difficult, but my view is that it's always best to choose the chart that fits the audience's expectations, even if it doesn't always fit the "correct" way. For example, say you want to show heart medication prescribed in a hospital, over time. The technically correct way would be a line chart, but if the audience is more focused on revenues and costs, a bar chart that shows each quarter, plus sections in each bar which department prescribed the drug, would technically be best.

Generally, keeping charts simple and easy to digest is advisable. However, if you do need to reach for more complex visualisation due to the nature of your data, do so sparingly and consistently, to improve reception. Nobody wants to spend 5 minutes to understand each single chart in a report before taking in the findings.

Para mim, três pontos s?o essenciais na visualiza??o de dados de pesquisa: 1) Validar os dados para garantir sua precis?o e relevancia; 2) Considerar o meio onde os dados ser?o exibidos (celular, computador, tel?o, coletiva de imprensa, jornal impresso ou online) para escolher gráficos e elementos visuais que atendam ao público. Por exemplo, preciso mostrar toda a série temporal no tel?o do evento ou apenas os números mais recentes? 3) Integrar a narrativa com a visualiza??o, utilizando os dados para contar uma história clara.

Consider adding a fourth step/go beyond the selection of visual elements. Equally important are captions that provide a clear and concise description of the visualization. Oftentimes, researchers verbally interpret the visualization for their audience. The visualization and the talk should be complementary; each supporting/enhancing the other. It might be helpful to expand on this point in this article.

Less is always more. It is usually better to have less on a chart than it is to have more elements, because you're trying to fit this in with a story more often than not. Here's an example. Labels are great, but if you have hundreds of possible types of labels (e.g., the dosages of drugs and brands being used to treat diabetes), that would clutter the chart. Instead, consider simplifying the graph to have less labels, and explain more in prose instead.

One key principle I always emphasize is that a well-designed visualization should be self-contained. This means that the visual elements, such as charts and graphs, should convey the main message and insights without relying heavily on explanatory text. The text should serve as supplementary information to support and enhance understanding but should not be the primary means of conveying the data's story.

Data visualization is the best way to present your research findings in a much more simplified maner to the target audience. The whole idea is simplicity, as such the illustrative elements selcted for use must fit the best use practices. For example, must fit the: audience, and purpose. The choice of the chats, colors, spacing, fonttype and size, headings must appropriately be used to communicate what the study findings are. The elements used in data visualization must not exaggerate but they enrich the presentation of findings to the audience to allow easy understanding of the data. Appropriately selecting and presenting the elements will also help reduce the noise in the visualization.

This step is under-emphasized and often overlooked, but it is the key to great visuals. In addition to doing your self-check or a standard QA, take it for a test run with your target audience. Show it to a few people and observe how they read it. How much time do they take? This helps you uncover blindspots that you may never discover with standard checks.

Your data visualisation should be as accurate as possible without leading to a loss of clarify. One key issue I've seen is the Y-axis being incorrect for the context. For example, I have seen bar charts which compare two groups of people for systolic blood pressure, which sits at 120 for healthy, and 160 for unhealthy. However, the creators of this sometimes start the bar chart's y axis at 0- which hides the differences between groups. Or in other circumstances, talking about sound levels with a linear y axis- but the correct way would be logarithmic. Talk to experts before presenting it, if unsure, and think about the audience too.