« Reading | Main | The meaning of most »



Really? They don't look like mirror images, even with the scale changed (in my head). Care to plot them on the same scale?


Awesome! I love how simply you broke this one down. Is there ever a time when dual-y axis plots are a good idea?

Floormaster Squeeze

Personally, I would always include the origin so you get a better absolute idea of change (there are exceptions based on audience but for a general audience this holds). You can make miniscule changes look big with scale. Playing around with scale on two axis means that I can make charts say nearly anything I want without any need for "reality" to play a role.


As it has been already noted the series are obviously not mirroring each other (one is constant in 2001-2004 while the other is growing). It's also easy to verify that they are not adding to 100%. You don't even need to look at the axes they're crossing at different levels at the beginning and at the end of the period, there is no way the sum could be constant. I guess you can still say the chart is evil because it doesn't start at zero or something... but it doesn't show percentages and it doesn't pretend to: "(in millions of units)".


Commenter #1 and #4: See my revised post. My point would be better made if the chart plotted proportions. But I hope you get the bigger message which is that the dual-axes chart is open to a lot of mischief.

Adam: I almost never use dual-y axis. The only time I'd use it is if the two data series have exactly the same scale (in terms of both the range and the units), e.g. two data series both of which are proportions. But then in those cases, why not use a scatter plot?

The comments to this entry are closed.

Kaiser Fung. Business analytics and data visualization expert. Author and Speaker.
Visit my website. Follow my Twitter. See my articles at Daily Beast, 538, HBR.

See my Youtube and Flickr.

Book Blog

Link to junkcharts

Graphics design by Amanda Lee

The Read

Keep in Touch

follow me on Twitter