Y? To Trick You, That's Why.

Fox News (and the Fox family in general) is somewhat notorious for manipulating the Y axis of their on-screen graphics to make misleading, if factually correct, arguments. These charts are the dark side of data visualization. 

From last week, this fun one:

Oh, that darned unlabeled Y-axis!  Here's what it looks like if the Y-axis starts at 0.

And here's what it looks like if we keep the same axis scale, but extend it downward to 0.

Here's another Fox gem, this time from Fox Business, dating from the summer of 2012.

Hmm...I sense a pattern! At least in this instance they were nice enough to label the Y-axis, even if they started it at 34 and used the right-hand edge of the screen. But what should it look like?

Doesn't seem quite so scary when you fix the Y-axis, now, does it?  It's so troublesome when accuracy gets in the way of a good story.

Here's one last one, from December 2011. We're leaving bar charts behind and moving into the world of line graphs, but with some Tufte-esque chart junk...and some other problems as well.

Look all the way over to the rightmost point. Oops! How did that 8.6% bubble up so high there?  Did you forget to Shift+Down Arrow that last data point when you cut and pasted it?  What's more, did you just randomly pick a Y-axis? Because it doesn't match your data points, even a little bit.  

(I suppose, at least if you do your axis that way, you get to write "10%" on the screen, so the implication is that double-digit unemployment is possible, or even likely. Subtle, clever, and devious.) 

But here, let me get that for you. Hope you don't mind if I clean up some of the junk:

That "corrected" chart is more accurate, but when you see that drop in unemployment from June to November, you lose the FOX-ish argument that Obama makes unemployment worse.  So how can we obfuscate that irritating data while not actually lying?

How about this: I could make the Y-axis go from 0 to 10%.  The upside is, it would at least get rid of that pesky decline in unemployment, and I get to put the number "10%" back on the screen...but the downside is that it would get rid of pretty much all the change over the year, in which case, what exactly are we trying to show here?


Point being, there are plenty of flat-out BAD graphs out there. But it's the misleading ones we should be on the lookout for...and should be careful not to use ourselves.


Mike Cisneros is on Twitter at @mikevizneros.