Not to scale

Check out the scale at the bottom: first it’s in months, then it’s in years, and then it just seems to vary arbitrarily. It also seems to be based on distressingly few data points (no doubt because it had to be created by hand — any self-respecting data visualization tool would have to be twisted and tortured to be made to introduce a distortion like that.)  Junk Charts dissects the problems further, and brings R to the rescue with a more reasonable visualization of the data, where the scale is uniform and introducing smoothing to better highlight the trends:

In comments, Rapp says “It is known that

Check out the scale at the bottom: first it’s in months, then it’s in years, and then it just seems to vary arbitrarily. It also seems to be based on distressingly few data points (no doubt because it had to be created by hand — any self-respecting data visualization tool would have to be twisted and tortured to be made to introduce a distortion like that.)  Junk Charts dissects the problems further, and brings R to the rescue with a more reasonable visualization of the data, where the scale is uniform and introducing smoothing to better highlight the trends:

In comments, Rapp says “It is known that this visualization is not perfect, but simplification can help a newspaper reader to get a quick comparative idea of the current market meltdown.” I can’t say I agree — that bizarrely warped scale only distorts the reader’s impression of the differences between these bear markets, rather than simplifying the interpretation.

Junk Charts: Don’t mess with the scale