A New Type of Chart for your Data Visualization Consideration: The Cartesian Waffle Chart
Let's call that the Cartesian Waffle Chart.
I can't find it in all the usual places you'd look, from just Googling "waffle charts" to Ann K. Emery's Depict Data Studio's Chart Chooser to Xenographics. I ever messaged Alberto Cairo about it and he drew a blank.
It's super weird to be claiming I invented a thing. I am a professor, and the idea of plagiarism is pretty horrifying to me. I posted it up on social media and a part of me keeps waiting for someone to message me with an "Actually...."
But that hasn't happened, yet, so I'm just going to plow ahead.
Waffles Good!
My graduate students in Data Visualization class know that I am a huge fan of waffle charts. I love them because they don't ask you to transform numbers into spherical areas (we are just fooling ourselves if we think most of those pie and bubble forms are processed by audiences correctly or in anything more complex than bigger and littler). They also don't ask you to transform numbers into shapes where convention tells us to ignore the shape part and just pay attention to the length, like columns or bars. And as we generate more and more complex chart types where various elements of other charts are mashed together (seriously, check out Xenographics!), calculations and conventions of areas are key trouble areas.
Graphics should help us understand! And maybe they can be complex and ask us to look at them for a minute to draw greater understanding than simple forms like they are number poetry. But in many cases clarity is the key, and so I value forms that reduce cognitive load.
Because waffle charts are just blocks, really, and we all remember stacking blocks at some point (don't we?), I think they have an immediate effect that other forms don't.
Waffles Bad!
The trouble with waffle charts is that not converting numbers into other forms means that there is a representational problem as the number of numbers increases. Take a look at this snapshot of a Google image search:
With multiple numbers, either we just make a lot of waffle charts or we stack multiple colors of blocks in one space. The first option becomes just decorative at that point. Don't you just look at the numbers and move on? In which case, why bother with a purely decorative visual? The second one is more interesting, but because there is no good way to figure out how to start and stop the different zones of color, we are often left to, again, just look at the number label and basically ignore the visual. Compare that to how much easier it is to contrast items on a treemap, for example.
When you try to add another variable, like I did in my chart, it just becomes impossible.
Putting Descartes before the Horse
I created this visual for use in a faculty meeting to help participants understand the results of a survey. I tried different forms and was unhappy until I went back to my old favorite, the waffle. But I wanted it to be more immediately obvious how big the different areas were. And only then did it really strike me that the waffle chart is so often just used with percentages and not real numbers. So escaping the boundaries of the multiplication table grid wasn't something that had occurred to me before.
I separated the box clusters in just horizontal spaces, but I wanted more comparison, so I needed the boxes closer together.
Thus, a version that borrows from the coordinate plane, which allows us to compare 2 variables really easily. And because it was visually simpler I felt like I could add the 3rd variable with the dots without losing clarity.
It worked really really really well in its moment. My point was clear and a number of faculty messaged me how much they liked the visual.
Onward!
What now? Well, try it out if you're that sort of person. You really have to draw this thing on Excel, though. That's not too hard, but in the age of Tableau and Power BI, maybe that's not your cup of tea anymore?
If you do try it, please let me know. I'm curious about whether or not it was helpful, but I am also interested in tracing innovation processes in the field of data visualization with my graduate students, and it would be useful to have a close-up view of that process.
I'm going to keep the idea percolating in my head for a bit and try it in other contexts. I'll report back later.
Thanks for reading, fellow nerds!
