2016 Postmortem
Related: About this forum****IBD/TPP Poll HRC+3****
http://www.investors.com/politics/clinton-extends-lead-over-trump-to-3-points-and-99-million-ibdtipp-poll/This has been the conservatives favorite poll for months. It has been promoted as the most accurate poll of the 2012 race.
Their new number today:
HRC 44
Trump 41
All of the polls are moving in the same direction with HRC leading from 3-7 points Nationally. HRC is in a commanding position heading into the last week of the campaign.
Farmgirl1961
(1,493 posts)But I still DO NOT LIKE seeing her numbers sliding down on 538.
Loki Liesmith
(4,602 posts)Farmgirl1961
(1,493 posts)but you're right, I should just ignore it all...but I'd like to know what the "real" polls are looking like and better understand why Nat's 538 in general seem so much lower than any other site.
Loki Liesmith
(4,602 posts)Because his predictions in the primary went so spectacularly wrong. He had trump <10% likelihod to win the nomination.
So he changed his model to react to every poll that comes in. Others don't so much.
LAS14
(13,749 posts)... flipped from light blue to light pink is Arizona. Would it account for the drop in % likelihood to win?
Loki Liesmith
(4,602 posts)All his states have correlations built into them. So if some states are strongly correlated to movements in MO, then their margins will slightly decrease too. This has a significant effect across many states.
Farmgirl1961
(1,493 posts)Any thoughts on how you think the 538 numbers will move (a little more toward Trump? a little more for Hillary?) based on state polling numbers?
Loki Liesmith
(4,602 posts)His model is too variable this year.
Farmgirl1961
(1,493 posts)so if/when it happens I won't be in total freak out mode! What about other models (i.e., Sam Wang)? Thoughts on where you think they might land?
Also -- any ideas on how early voting may or may not impact the models? When state polls are gathered, at this point, are they still going for "who are you likely to vote for?" versus "Have you voted and if so, who have you voted for?"
DCBob
(24,689 posts)I think they are finally revising their biased methodology so they fit closer to the real numbers.
aaaaaa5a
(4,667 posts)Assessing polls by their final numbers only is BS.
If your polls were off for the entire election and the at the last second they make a "market adjustment" so they can be credited with scoring the race correctly well... that shouldn't count.
I've noticed over the last 2 Presidential cycles, conservative polls do this frequently.
There is no way this poll was a correct analysis of the race, and now at the end, they are lining up to get another good rating.
Those ratings should somehow include polling performance during campaign.
aaaaaa5a
(4,667 posts)Loki Liesmith
(4,602 posts)usually this works by polling companies getting skittish about releasing numbers they think are outliers. With tracking polls, that's difficult because they have to release one every day.
Blue Idaho
(4,988 posts)To cover the Republican bias of that poll.
Roland99
(53,342 posts)This reversal is huge and only been a few days