I still don’t know how they can expect to poll accurately at all. The majority of people under 40 either don’t have a landline at all, or they don’t answer any calls from numbers they don’t recognize. I fall into this group, and thus have never been polled.
Caller ID does introduce a wrinkle. But I’d guess people who wouldn’t be willing to answer polls in the first place and people who don’t answer unknown numbers are mostly the same people. 538 also grades each firm by analyzing the historical accuracy of previous polls, so firms with a better track record of reaching these people are weighted accordingly.
How can anyone make a firm prediction with 20% undecided/third party? Most Dopers who have expressed an opinion on Johnson say he’ll lose a lot of his voters by election day. Okay, so where do they go? Is Clinton really safe with even a 5 point lead and Johnson with 5 points to give away plus the 10% purely undecided? With those numbers her chances would have to be rated 75% or so even on election day, much less 60 days out.
At 2-3 points ahead, Silver correctly has her downgraded to closer to 70%.
Why is everyone treating Silver’s methology as so mysterious? He’s explained it in the past. He does indeed have a lot of demographic factors he has statistics on, ranging from the obvious like party identification and race to more subtle ones like proportions of various religions and employment sectors. He then runs statistical analyses using past election results to see how significant each of those demographic factors are (IIRC, for instance, proportion of the population that identifies as Southern Baptist turns out to be fairly significant), and then uses those significances and the demographic statistics to construct his covariance model.
Oh yeah, Silver’s got her down to 66.4% in polls plus.
Ah, the old “because cell phones” argument.
People trot this out every election. They said in in 2008 and they were wrong, and in 2010 and they were wrong, and in 2012 and they were wrong. They say it here in Canada when elections roll around and they’re wrong. It is always wrong; the polls turn out to be right. Pollsters are not idiots, and are aware of the existence of cellphones.
Would that include my parents? They’re above 40, but don’t use a landline- they’re deaf and they use VRS (Video Relay Service) to communicate via VoIP. (It’s handy). I bet you that even TTY users aren’t being asked either (who still uses TTY?)
Today’s polls show her falling a little more: +3 in two way, +2.1 in four way.
NOW has she bottomed?
Who knows? Day to day fluctuations don’t matter. We’ll know more over the next few weeks, and after the debate.
That sounds like you are acknowledging uncertainty, when many posters here think Silver is giving Trump better odds than he has in reality. Whereas Wang and a couple of others are basically saying, “Clinton wins barring the apocalypse.”
Silver has a better track record, but I’m not going to panic when she’s gone from a 4 to 1 advantage to 2 to 1. She’s still significantly ahead, along with money and ground game advantages that aren’t measured by polls. I’m not comfortable, and I won’t be until the day after, but right now I’m still feeling pretty good.
I agree with that. She’s still more likely to win than not(66-34 odds seem quite reasonable at this point), and she can help herself with a good debate performance. Or a friendlier media, which she can help herself with by being more open to the traveling press corps. But the Michael Dukakis strategy is a loser. It’s not a week before the election, it’s two months. And as a recent article from 538 points out, if Trump can make this election seem normal, he’s in good shape. All they have to do is keep him from saying anything dumb for two more months and they may pull this off.
I’m pretty confident that the probability of Trump not saying anything dumb for two months is zero.
Nate Silver and PredictIt are both around 70-30 right now. That sounds about right.
The Commander in Chief Forum is in about three hours. I wonder if that will move the needles.
Oh, is that all? Well, nothin’ but blue skies and sunshine ahead!
Does anyone still want to take issue with Nate Silver’s probabilities?
Depends on how long this lasts. But yes, offered a bet at Silver’s current polls-only or polls-plus odds (3:2) I’d take the bet in a flash.
I’ve already got my money on Clinton at 59 cents a share, but that’s not really the issue with this thread. Many posters were thinking her odds should be above 75%, even 90%. Now that her lead has shrunk to 1 point in the 4-way and just under 2 in the 2-way, I’m not sure that any model could give her such high odds, although Wang is a stubborn SOB and has her above 75%.
Wang stays true to his transparent model which relies on state polls. He explicitly states it is slow to respond as a result. That makes it less noisy but also less quick. Current posts acknowledge that it will move as state polls come in.
Yup. Still leaning towards Wang’s/PEC probabilities. He’s downgraded a little too with the new polls. 77%/86% chance for Clinton win. If the swing states look bad end of Sept I’ll get nervous.
What gives me less confidence in Wang this week is that the actual movement of his EVs for the few weeks toward Trump is well outside the width of his confidence intervals going forward (here). If what we’ve just witnessed is outside his CI, how can his CI be taken seriously?
By that simple gut check, current point estimate may be fine though slow-moving, but his Nov win probabilities would be way off.