another 3 bad charts-01-01

NSFW: CHARTS you shouldn’t be looking at

Today is bad chart day. That’s an unofficial, non-hallmark holiday. It happens today and every day in presentation land. Bad charts, ones that mislead, confuse, and generally make your audience’s eyes glaze over decorate millions of presentations all over the world. In this occasional series, I’ve already spoken about the Timelord, (for those of you that live in alternate timestreams) the Harlequin (for the motley fools obsessed with color) and the Lazy Susan (for the people that like to make people take the long way round). We’ve covered the Pointless, (because you really don’t have a point, just some numbers), the Martha Stewart (for those of you into over-decoration) and the Australian (because you like your numbers backwards).

Here are 3 more. Check your decks to see if you are making any of these mistakes.

The Lyin’ Y

Time-Lord Chart_The Lyin' Y

Your smile is a thin disguise. Showing a chart with a huge uptick in ______ (you fill in the blank — growth, revenue, share, customer satisfaction) and being happy about it may be a trick of forced perspective. Foreshortening the Y axis on a chart, by starting at a number other than 0, or picking a very small range to highlight small variations is playing with the truth, and a strict no-no.

We had a recent experience with a client that had a big uptick in customer experience numbers, but no corresponding change in churn. The reason? a foreshortened Y axis on their Cx score. When you’re creating your next set of charts, and you have a number range between 0 and 100%, do yourself a favor and use the whole range.

The South Beach Pie

Time-Lord Chart_The South Beach Pie

The problem with pie charts, according to Cleveland and McGill in their paper Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods, humans are a poor judge of angle, therefore spotting the difference between a 17% share of the pie and a 20% share of the pie is near impossible. Labels and mouseovers, having to read data rather than see data, just adds another layer of work to the audience — something you don’t want to do.

The South Beach Pie is one of the worst examples of a pie chart. It has so many datapoints that the pie becomes very thinly sliced. If you must use a pie chart, (and there are many better alternatives) don’t use one when you have more than 3-4 data points.

The Delicious Donut

Time-Lord Chart_The Delicious Donut

Loved by the graphic design community and reviled by the data visualization community, the Delicious Donut is very popular in business dashboards. It adds that little bit of extra sizzle when you only have one datapoint, but need to draw the eye to it. E.g. 25% of people like sprinkles on their donut. Boring as a sentence, cute as a visual. That’s why graphic designers love it. It’s a visual ornament for a single data point. And that’s why the data visualization community hates it. It’s a single data point. (And also the aforementioned Cleveland and McGill angle problem.)

I personally use donut charts, and I think they have a place, for precisely the graphic design reason stated above, sometimes you only have a single data point and need to make it stand out.

Have you made these mistakes?

You may have made some of these mistakes. There are some great experts out there that can help. If you want to improve how you handle data, polish your plotting or brush up your bar charts, I recommend you start here.

Gavin_Animated-GifGavin is a founding partner at fassforward consulting group. He blogs about PowerPoint, Presenting, Communication and Message Discipline at makeapowerfulpoint.com. You can follow him on twitter @powerfulpoint.

More at Google+Facebook and Pinterest. Comments are welcome, links are appreciated. If you’re interested in writing guest posts for this blog, please contact me.

2 Comments

  1. adam | May 15, 2014 at 2:30 pm

    Thanks Gavin for a great post. Question about the Lyin’ Y … what if you were representing a data set for which there is rarely any fluctuation greater than 1 or 2 percent AND that 1 or 2% difference was actually very consequential. For eg., global average temperatures, where a few degrees can be a profound change. You think it’s okay to ‘zoom in’ on the Y axis at that point or just use a different chart to represent the data?

    • Gavin | May 15, 2014 at 2:50 pm

      Hi Adam,

      Thanks for the question. I do think it’s OK to zoom in – that’s always an option. Another thought may be to look to show data that shows how profound a change there is in a few degrees. For example, what does a few degrees difference make when you think about crop production? In your case, I think it’s about making the case around what difference a degree makes, not what difference a percentage makes. An example from a subject I am more familiar with – Churning 0.1% of your customer base is (or seems to be) inconsequential. Translating that to real impact, e.g. we lost 2,432 customers yesterday, makes the number seem more real. See the post I wrote Comparisons Speak Louder Than Words.
      http://makeapowerfulpoint.com/2014/02/13/comparisons-speak-louder-than-words/

Leave a Reply