WTF Visualizations names and shames terrible data viz

The data are that 13.2% of non-hispanic black women have preterm births, 9.0% of hispanic women, and 8.9% of non-hispanic white women. If you had equal numbers of women from each group, 42% of the preterm births would be from non-hispanic black women. That 42% is exactly what the wedge shows.

The reason for presenting the data this way is even clearer with the next chart from the source article:

From the numbers alone, or even a bar chart, it would not be as easy to immediately see the conclusion: black babies are nearly twice as likely to be low birth-weight than the other two groups. If you had equal numbers of babies from each group, there would be nearly as many low-birth weight black babies as from the other two groups combined.

I feel like the choice of representation clearly made this conclusion stick out. On reading the article, I saw that this was, indeed, the point of the article. So it fit.

3 Likes

That leaves 68.9% of preterm births unaccounted for.

Does it not mean that 13.2% of non-hispanic black women’s births are preterm etc…?

Gah, word-scramble, not a confusion in my head. I was rushing forward in explaining the second-half while still writing the first half. Corrected.

1 Like

Where’s that music history chart?

2 Likes

Interesting that the first graphic uses the Black Power symbol for the Violent Offenses category…
Telling?


5 Likes

Telling of what? That there are only so many ways to draw a closed fist. The simplistic design is the same as the other icons, and not exact to the Black Power logo. Given that many if not most “violent” offenses are assault, that is an appropriate icon.

2 Likes

Absolutely, the man knows his infographics.

Slightly O/T: About a dozen coworkers and I attended his one day course (with hundreds of strangers, of course) on presenting data and information. It was awful. Turns out knowing how to present data visually for someone to ponder on their own has essentially nothing to do with being able to present it well in a talk, to deliver a talk in a way that holds attention or at a good pace, really disappointing. Glad my company paid for it and not me.

2 Likes

Unnecessary gears and levers doing things gears a levers can’t do? This goes way beyond venial and dives deep into psychological torture to anyone with a basic understanding of mechanics.

3 Likes

I love how Brass Eye satirized bad data visualization.

Sad thing is that sometimes I find genuine examples which are equally perplexing.

4 Likes

Meh, poor data viz is poor data viz all the way down.
Sure, a fist is an appropriate icon. But don’t tell me that you think all the iterations of similar looking fist graphics aren’t based on (what I presume to be) the original design. It’s a good design, I’m sure that is why it has been copied so much.
Coulda used this one that kind of looks like two bunnies getting ready to kiss…

4 Likes

Aw man, you got you peanut butter in my chocolate.

2 Likes

As with all things graphic design, it entirely depends on your audience. Given that this infographic appears to be about SAAS for B2B marketing, and that the number of B2B marketers (the audience) who are also engineers is nonzero, it’s criminally bad design. A data visualization should not be designed to cause physical or emotional trauma to any member of the intended audience.

1 Like

This is a little embarrassing to admit, but I can’t quite work out how this site works. It says “For a discussion of what is wrong with a particular visualization, tweet at us @WTFViz”. Meaning what exactly? Send them a DM and they’ll reply with a discussion?

1 Like

I can’t even describe how awful nearly every chart on that horribly laid out article is. Not only did they shade the charts by skin color, they only did it about 90% of the time so you literally have to reference the legend on every fucking chart. On top of that they use the “dark” color as the “bad” color on their national maps. There is no possible way CNN could have done a worse job presenting this data, and I don’t see how you think it gives any meaningful data - even with comparing chart to chart.

And yes, a single bar chart could have been used in both of those bar charts and been universally more effective in every single way. Here’s is literally the automatic chart excel shits out, notice how much better the information is presented for comparing the statistics.

1 Like

Should be a stacked bar with labeled segments, I think.

fundamental(ly) even

This topic was automatically closed after 5 days. New replies are no longer allowed.