Wednesday, July 20, 2016

Two Three polls move large...



Democracy Corps 
Hillary was up 11
Now up 3


Ipsos/Reuters
Hillary was up 10
Now up 5

ARG
Hillary was up 9
Now up 3

4 comments:

Anonymous said...



but when you put on your beer goggles, hillary's lead is growing.


Indy Voter said...

You left off Rasmussen. Oh wait, that one moved large towards Clinton.

wphamilton said...

I wish I had the data to explore it, but I believe that the idiot demographic, necessarily choosing for idiotic reasons, is distributed randomly between two idiots. Those whose votes are determined by party affiliation may be included in that number, and it is my belief that idiots affiliate with parties also based on idiotic reasons leading to essentially random probability distributions, with variances due to multi-generational positive feedback and community bias.

Therefore we have the actual selections made by those of at least the first standard deviation beyond the median, of knowledge and intellectual capacity. I'm probably being too generous and it's more likely closer to two standard deviations. The upshot of all that, if I'm right, is that small movements in the polls (if truly reflective of the electorate) are in reality much more significant than they appear. Chop off than random distribution part, say 40%, and look at the remainder. A swing of 5-8 percentage points is huge.

But we all know that small poll movements are not truly reflective of the electorate, and one big reason for that is we have no reliable way to decide who really IS likely to vote, not with the precision necessary to reflect the non-random impact of that 15% non-idiot one sigma demographic. This kind of uncertainty is possibly why Nate Silver came up with his "Six Hurdles" for Trump to overcome (with no justification other than it seemed so to Nate), assign each of them a 50% chance for Trump (with no justification other than it sounds like a good number) coming to his 2% chance for Trump getting the nomination. Silver was so badly off the mark that the real question is why would any mathematician construct such bizarre reasoning in the first place. He knows better, has to or he'd never have had any success in 2012 predictions. I think this is the answer. We just don't have reliable polls for this cycle, since "likely voters" is unstable, mixed up by Trump himself and a little by Sanders as well. So Silver at the FiveThirtyEight blog has been relying on what we professionals call the "wild ass guess" or WAG for short.

My own WAG, based on the same polling problems, goes in the opposite direction. I think that Trump's eventual vote is under-represented in these samples, and he has at least a 50% chance of winning at this point. Probably more, but like Nate Silver I can't quite make myself predict a Trump victory in public. Yet, there it is ...

Indy Voter said...

Polls taken today can't really predict what might happen in the next 15 weeks or so. Dukakis riding in a tank while Bush wrapped himself in a flag changed the 1988 race. McCain's solution to the Lehman Brothers collapse, "Fire Cox!" ended whatever chances he still had. Reagan's second debate in 1980 propelled him to a landslide.

Clinton has shown herself to be cautious far too often. That may hurt her big time when it comes to fending off the outrageous attacks Trump is sure to make. But the flip side is that Trump's outrageousness could backfire on him.