I'm either spoiled by Tufte's work, or just not getting it. What dimensionality is the spiral capturing? I get that it looks neat, but in terms of helping me to understand what the data is telling me I just don't get it.
This isn't sarcasm, please explain what I'm missing.
Edit: This work is quite solid by the Tufte principles that I can see how to apply, such as using ink for data and not decoration.
There are little '+' hovers on the image. I'm assuming you missed them.
> Observe how the data visualization starts out like a normal line graph and then sharply tangents off in a series of colors that end in a generous spiral of red. "It literally looks like a piece of Modernist art," says Rusert.
> The art of the data visualizations is partially what made the information so compelling. "We've now had it confirmed by a couple of different art historians that it seems like [Du Bois' team] were also developing novel and unique styles in order to grapple with and represent this data," says Rusert. Take the spiral design, she says. "it raises questions about why they were departing from these more geometric shapes in terms of representing data."
> To Rusert, there's a lot of intent that can be read within the way the information is translated through the drop of the red spiral. "It sort of a reminds me of lynching and forms of racist violence that were happening during the period," she says. "It also reflects something about just the surreal nature of segregation itself, and what it meant for African-Americans to be living within this kind of regime that was wholly irrational and was also backward looking not forward looking."
These works are meant as much to be artistic and emotional communication as they are a conveyance of numerical data. The image of the African American population spiraled in on itself in blood red is pretty striking.
I guess I should preface this by saying that I'm not being particularly trained in this; I've skimmed Tufte, but that's it.
To me, the interesting thing about it is that it captures both the utilization of two dimensions of the paper/screen and the representation of a one-dimensional quantity. Humans are notoriously bad at intuitively translating among quantities of different dimensions, e.g. translating the area of a circle to a population.
I got that, but why not just a bar chart? Is there something I should be learning from this design to use when I share information professionally? Because right now if I were to chart data in this format for work I'd raise a lot of eyebrows, and I'd like to know what the advantage is.
(For context: not a dataviz expert by any means, but I've done a bunch of work in that space for the last several years.)
For sharing information professionally? Not much - but you have to remember the context. Du Bois and his team were preparing these visualizations to be shown at the World Fair in Paris, not for a business meeting. As important as the data they portray is their underlying goal: to show that African-Americans are the equals of their European / Caucasian counterparts in cultural and intellectual endeavors. Presenting the ongoing socioeconomic plight of African-Americans as Modernist data art - at a time when Modernism itself was only just starting to spill over into the visual arts - definitely supports that goal.
Now, on to the visualization merits of the spiral. We have four data points spanning two orders of magnitude. To show those on a bar chart, you'd have to either: a) use a logarithmic scale, b) use a linear scale and accept that three of the bars are tiny, or c) use "scale breaks". Instead, Du Bois presents the data in a way that draws attention to the most prominent data point, while still preserving linear scale.
There are some highly effective techniques in the other visualizations as well. The property valuation chart uses annotations to provide narrative context. The freemen / slaves proportion chart uses filled-in space to lend it a stark visual weight that, again, speaks to the narrative impact. Data visualization is storytelling with data, and these charts are clearly designed to get their narrative point across - quite effectively, I'd say.
Finally: Tufte was printed in the 80s, nearly a full century after these visualizations. That, more than anything, makes these striking. With the exception of Playfair's charts, Oliver Byrne's treatment of Euclid's Elements, certain railroad timetables, and few others, the world hadn't seen much in the way of truly effective visualization. They were literally inventing effective visualization techniques. That we don't hear more about Du Bois' charts is unfortunate.
A linear bar chart would have the red bar be 10-20x as long as the other bars. That is a "correct" visualization but looks visually quite different from what was conveyed here.
No, you probably shouldn't be learning from this design; it requires significant skill and imagination to deploy interpretive graphics like this. Stick with simple things.
This isn't sarcasm, please explain what I'm missing.
Edit: This work is quite solid by the Tufte principles that I can see how to apply, such as using ink for data and not decoration.