Wow! In general, charts that have two axises to display trends when there are two different units are hard enough to decipher (temperature and humidity for example). But when the unit is the same (% usage), to use two different scales is just insane.

My suggestion: They should change the chart to just have one axis, and have the chart be "% change in usage year over year".

This is roughly what the graph looks like if you change the axis to be % change year over year: http://imgur.com/a/H6T5d

Could you do a chart with cumulative change over those years?

Is this what you're looking for? http://imgur.com/a/IGJ2h

It's stacked, not cumulative. Cumulative would be two cumulative running sums (for beer and pot) from the beginning of the period.

Not only are the labels shifted by more than 50 percentage points, the scales are slightly different too, meaning not even the slope of the lines can be compared.

"How to lie while telling the truth, with figures"

Yes, and it omits both zero axis points. The movements are just a couple percentages but look huge setup that way.

The way they do all stock charts. Because otherwise CNBC would be really boring. OMG The meaningless Dow Jones Index jumped 0.25% !!!

Yes, I agree. Bloomberg is even worse when they show these giant charts with unreadable y-axis's.

That's the great thing about statistics! You can manipulate them to support any point.

That's the great thing about misrepresentation! You can manipulate any representation to prove any point.