Skip to content

Close Reading Data Visualizations

As is often the case with such TED talks, I watched Gary Flake’s demonstration of Pivot with a mixture of awe and jealousy. I want that kind of thing for the deep web as well as the free web!

Go watch it, then come back.

No, really. I’m about to reference a specific visualization, so you should see it first. If you get bored, just watch the first sequence (until the flying people-graphs are done).

Ok. Wasn’t that cool?

The instructor in me, though, noticed an implicit message in the visualizations that I think would reinforce incorrect assumptions that my students make all the time. My students are constantly looking at census data, for example, and hoping that they can make talk about how many of these people from this chart in their hands that describe educational levels — how many of these people died in that other chart on accidental deaths. They’re wanting to track individuals rather than talk about probabilities and percentages. And the initial example that Flake uses to talk about mortality and age absolutely reinforces that faulty understanding of the data. Icon-people fly from one column to the next as he filters for different characteristics, making it seem like if you just concentrated enough, you’d know everything there was to know about that one blue guy who started off 3rd from the right of the 4th row.

I really wish the visualizations had figured out a way to make each one appear to be exactly what it was: a snapshot of a sample. Right now it looks like they’re drawing on incredibly detailed longitudinal data.

Published inRandom Thoughts


  1. Megan Megan

    My challenge with that graphic was related to the gender icons. (And not the gender-binary aspect of the blue/pink characters, although that chapped my hide as well.) What I mean is that you have a column that is 4 figures wide by some number of figures tall. The “boy” figures were always on the bottom, so you could easily judge boy figures against boy figures without even paying attention to the “girl” figures. To be able to visualize the girl figures against the girl figures, you’ve got to drill down one more level. I wish it was possible to transpose the boy/girl figures on those graphs, which would allow you to easily compare girls v. girls or boys v. boys, all while still visualizing the ratio between girls and boys.

Comments are closed.