Friday, February 05, 2016

Crappy Statistics

Vox in talking about the liberal/moderate/conservative line and how it helps Clinton presents this chart:
This is an example of incomplete data that doesn't actually help inform.  If more conservative people call Romney liberal, Obama very liberal, and themselves very conservative, but more liberal people call Romney conservative, Obama moderate and themselves between moderate and liberal you can end up with that plot even if far more people are Obama supporters than Romney.

The perception of self vs. others is important and it is likely different between liberal and conservative individuals.  This plot is only useful if more liberal individuals and more conservative individuals have the exact same relative perception.  Nothing about that chart would imply they do.  

Obama is, on a historical scale, pretty much the same as Regan, which, at the most liberal, would put him right down the middle.  Romney, as governor would have been similar, though Romney the presidental candidate ran as closer to very conservative than moderate.  

Anyone using that chart to draw general conclusions is either lazy or bad at statistics.  Those numbers don't mean what they look like they mean.

No comments: