IS A PICTURE ALWAYS WORTH A THOUSAND WORDS?
Uncover the secrets of what makes some visualizations more effective than others
Data visualization helps us transform data to information by helping us see trends and patterns and draw insights. We have often heard that “a picture is worth a thousand words,” but what if the visualization is not created well? The way elements in a visualization are arranged can impact the way we perceive it, what we pay attention to, what we understand, and what we draw conclusions out of. In this article, we study a research paper that highlights the importance of effective data visualization. It throws light on how the arrangement and placement of data in a visualization can impact our understanding. This article will help designers learn how to create impactful visualizations.
Background and Past Work
Research work in the past has focussed on different visualization types such as bar graphs, scatter plots, and line graphs and studied the strengths and weaknesses of each. However do we know how arrangement of elements within a bar graph impacts the inferences we draw from it? The present study aims to explore exactly this!
The Great Epiphany
Bar charts are the most commonly used data visualization technique to tell a story through a complex dataset. But did you know that the arrangement of the bars can significantly impact the insights you draw from it? Not all bar charts tell the same story!
How the arrangement of data matters
A research study showed that the spatial arrangement of the elements in a chart play a key role in how the information is conveyed and understood. In this experimental research study, researchers showed participants four different visual arrangements of bar charts i.e. adjacent, overlaid, stacked, and vertical. The goal was to understand how participants compared the data and how they reached conclusions. They found that the spatial alignment of the data and proximity to each other matter. When elements (bars) are in proximity of each other, participants can compare them better. This is an important insight for designers and researchers who choose to represent their data through bar graphs. Through their study we learnt that not all kinds of bar graph presentations would yield the same insights and conclusions. Thus, in order for the readers to draw the right learnings from a bar graph diagram, it needs to be designed in a certain way.
How did the researchers go about finding this out?
The researchers conducted two experiments. In the first experiment, they asked crowdsourced participants to study 4 different types of bar graphs and report their takeaways. Researchers then analysed how the participants compared the bars. There could be 12 different types of comparisons participants made. (Cheat Code: Look at the picture below and follow the coloured element in each chart to understand the type of comparison) For eg. In chart 1, element 1 in group A gets compared to element 1 in group B, making it a ‘within element, across group’ style of comparison. In chart 6, two elements in group A get compared to two different elements in group B, making it a ‘across element, across group’ comparison. In chart 11, notice how all three elements within group A get compared to one another and all elements in group B get compared to each other. That is an ‘across element, within group’ kind of comparison. There are 12 such combinations as seen in the picture below.
Researchers wondered do participants compare the same element across different groups (Within Element, Across Group), or do they compare different elements across groups (Across Elements, Across Groups), or do they compare different elements in the same group (Across Element, Within Group)? Researchers found that participants most frequently made Within Element, Across Group comparisons, followed by Across Elements, Within Group comparisons.
In the second experiment researchers had data visualization experts choose the best visualization that they felt makes the right comparison between four arrangements through multiple choice questions. The study showed that there were some alignments between their intuitions and empirical results. For example, most experts agreed that overlaid information was the best arrangement, but did not reach a consensus over the other three forms of bar arrangements.
Through this study, researchers established guidelines for designers to choose the more effective data visualization technique for the right conclusions to be drawn. The findings were summarized as follows:
What does that teach us?
Understanding the art of visualization goes beyond choosing the right chart type. Understanding how to craft the chart is also critical since it helps the readers comprehend, process, and interpret the data in an insightful manner.
Thus, the method of unlocking the secret behind why some visualizations work better than others, is to understand the art of spatial data arrangement.
Concluding Thoughts
This research study shows how the arrangement and organization of elements in a chart can impact how viewers draw comparisons and make sense of the information. This is especially important for students, designers, and anyone who works with data. Empirical research and evidence is crucial in understanding the effectiveness of visualization, and one cannot only trust the judgment of experts. This research does pioneering work in making data visualization processes more effective and efficient.
Reference: Xiong, C., Setlur, V., Bach, B., Koh, E., Lin, K., & Franconeri, S. (2022). Visual Arrangements of Bar charts influence comparisons in viewer takeaways. IEEE Transactions on Visualization and Computer Graphics, 28(1), 955–965. https://doi.org/10.1109/tvcg.2021.3114823
