Numbers 2

I chopped up some numbers and rolled them into a graph.  As the numbers themselvesa re not actually important, I have left them off the graph.  Why?  One set of numbers has been churned to accentuate differences, so the actual value does not answer an intelligent question.  It is only meaningful in relation to the rest of the numbers, and we can see that relaton on the graph.  Hence no numbers.  The other set is simple enough, but is presented log-log, which again accentuates differences, etc.

The vertical scale of the graph is literally the most recent RCP average, but presented in log form.  The horizontal scale is a function of all of the recorded RCP poll scores, so that long-term performance matters more than (but not exclusive of) current performance.  So I think of the vertical scale as current performance, and the horizontal scale as staying power, in an already-demonstrated sense.  This assessment makes no claim as to future staying power, just what has been seen.

The diagonal line in the middle is a power-law regression, appropriate to the log-log view, dependent only upon the data.  That should represent a measure of who is currently overperforming or underperforming relative to expectation based upon prior performance.  I added a couple of blue lines parallel to that, one through Perry and one through Trump.  Finally, I drew an oval around a group I call the main body.

Using this over/under performance “hunch”, I wind up with perhaps some insight, and definitely some new confusion.  Why did Perry drop out but not, say, Graham?  Perry is underperforming badly.  Why is Fiorina surging despite current popularity similar to Huckabee?  She is overperforming, to Trump-ian levels.