Thread by @jburnmurdoch

@LyricalFalls Sorry but this take is so tiresome.

In dataviz, painting as honest & clear a picture as possible is about showing as much context as poss, not a zero y-axis.

Here, 20 years of context are carefully built up to show typical range, then we see how 2020-21 has diverged from that.

@LyricalFalls Extending the y-axis to zero would make the picture less clear. That’s all it would do. No added clarity. No added information.

@LyricalFalls Every tiny design detail in this chart is about maximising information and minimising risk of misinterpretation.

Gradually showing the context, changing the lower extent of the chart to make clear it’s not a hard floor, using light gridlines instead of a dark baseline.

@LyricalFalls All "but y-axis" tells me is someone has learned the letter of the law, but never really thought about the spirit of the law (making informative charts that don’t mislead people).

@LyricalFalls What’s more likely: that I just missed that day of dataviz school, or that I’ve been making charts for mass audiences for 10 years and discussing issues like this throughout?

https://twitter.com/jburnmurdoch/status/1176261830198648832

@LyricalFalls Charts like this always elicit a scattering of "but the y-axis" replies, but literally never result in anyone saying "wow {variable} fell to zero." It’s like a dataviz version of concern-trolling.

@LyricalFalls Sorry about the rant — it was prompted by the gradual straw-on-a-camel’s- back build-up of many many instances of this, not by yours alone! 😅

I just find it a really elitist take: "I a scholar of dataviz, think this could mislead someone", despite nobody being misled by it.