A new data visualization research paper finds that chart junk does not harm accuracy and actually improves recall. The paper is an interesting read but, unfortunately, not for the right reasons.
I’ll discuss the paper in an upcoming post. Today I just want to comment a sentence from the introduction:
“This minimalist [Tufte’s] perspective advocates plain and simple charts that maximize the proportion of data-ink – or the ink in the chart used to represent data.”
Now, let’s see what the authors believe to be a minimalist chart:
This is not a minimalist chart. It may be a plain and simple chart, but it’s also an ugly one. No wonder no one remembers it.
You see, when you strip a chart down to its basic features, a real minimalist chart also plays with variables like color to emphasize / de-emphasize the remaining features (series, grid lines, labels) to create an aesthetically pleasing experience that drives an emotional response. And that’s what the authors themselves do when reporting the results:
This is a fairly Tufte-compliant chart…
Problem is, if you misrepresent a concept it undermines the entire research. Instead of “minimalist” charts, the authors should call them what they really are: placebo charts. So, they should say: “In our comparison between junk charts with a low number of data points and placebo charts, the results suggest that…”
A note about data-ink ratio. I believe that the deep meaning of the data-ink ratio is not to remove junk, but to create more complex and insightful charts. Once you remove useless “features” you get more chart real estate and you can add more data (Tufte says: “to clarify, add detail”). And that’s how you really maximize the data-ink ratio, not just because you remove junk.
So, what do you thing? Are minimalist chart just some boring plain & simple charts or there is more about them than these authors are willing to accept?