2016 Postmortem
Related: About this forumWhat the likely voter models are missing: Obama's -- now Hillary's -- ground game.
This was an article from 2012 about Obama -- but Hillary's organization is on track to match his, if we all do our parts and never give up!
Notice that this article was written mid-October, and they had just switched to likely voters. I wonder why the pollsters have switched earlier this year.
http://www.thepeoplesview.net/main/epeoplesview.net/2012/10/ground-game-why-registered-voter.html
The media and the pollsters, as if right on queue, have started feeding the masses the "news" that the "race for president is tightening." I have made the case that most of that is horse manure, but I have also made the case in the same breath that the universe of who votes will determine who wins this election. In that vein, let's talk a little bit about the polling universes. Gallup recently switched to hilighting "likely voter" numbers rather than registered voters, but today's numbers from Gallup shows a tie on likely voters while the president is ahead 50-45 nationally among registered voters. A CNN/ORC poll from Ohio put Mitt Romney within 4 points of the President (51-47) among who they consider "likely" voters, whereas the same poll's registered voter preferences showed the president with a commanding 10-point (53-43) lead.
So, why does that matter? Aren't likely voters what should be looked at this late in the game? Not exactly. Firstly, the newly registered are almost never considered 'likely', since they have no voting history (it's like having no credit). Given the way pollsters determine who is a likely voter - which includes a given person's voting history - the "likely" voter universe is always skewed towards conservatives.
SNIP
I would contend that in this election, the registered voter numbers are more important than usual. Because in this election, pollsters are more likely to discount Democratic enthusiasm as a part of their discounting of 'unlikely' voters. In this election, pollsters are not likely to pick up on the minority, youth and poor groundswell (overlay this with women - especially young women) with their predictive models. In this election, the Obama campaign is getting those voters out. People are already voting in many battle ground states. In Iowa, four times as many Democrats requested vote-by-mail ballots as Republicans, for example. The Obama campaign has pushed the same thing in Florida. The President was getting the early vote out in Ohio himself yesterday.
The media has written and talked about Obama's doomsday before. People have bet against this president before. Exactly zero of those bets ever paid off. Get. Out. The Vote!
NewJeffCT
(56,828 posts)is that while Romney's ground game was not nearly as good as Obama's game, he at least had one. Trump's ground game is weak and pathetic, like the candidate himself.
Cicada
(4,533 posts)Obama won ohio by 3, close to likely voter poll. Not close to registered voter measurement. I guess we should look at average likely voter spread, average registered voter spread, actual vote for several elections.
pnwmom
(108,980 posts)uponit7771
(90,347 posts)Cicada
(4,533 posts)See fivethirtyeight.com/features/registered-voter-polls-will-usually-overrated-democrats/
That provides more general info on accuracy of registered voter and likely voter polls
I hope the link is correct. I have not yet figured out how to cut and paste on the tablet I am using
uponit7771
(90,347 posts)Fast Walker 52
(7,723 posts)good discussion of it with Mitch Stewart on the Sept 9th show here:
https://theringer.com/keepin-it-1600-podcast-politics-election-jon-favreau-dan-pfeiffer-220924af4c94#.roa8m9s51
And I keep thinking the models just aren't taking this into account.