Election Update: National Polls Show Clinton’s Lead Stabilizing — State Polls, Not So Much
Source: FiveThirtyEight
Were a couple of days removed from the point when almost every poll showed Hillary Clinton on a downward trajectory. Instead, polls over the past 24 hours have been more equivocal. National polls tend to suggest that Donald Trumps momentum has halted, and that Clinton may even be regaining ground. But Trump is getting his share of good results in state polls, which both show competitive races in some of Clintons firewall states and favorable trend lines for Trump.
Starting with those national polls: Clinton has regained ground over the past couple days in the national tracking polls conducted by ABC News and the Washington Post, The New Orleans Times-Picayune and the Los Angeles Times (although the LA Times poll continues to show Trump ahead, as it usually does). There arent really any national tracking polls that are still moving toward Trump at the moment, in fact, although some like the IBD/TIPP tracking poll show a steady race. Were notably lacking in the major, one-off national polls conducted by news organizations such as CNN or NBC News. Those should begin to be released over the weekend, and theyll have a lot of influence on the forecast. For the time being, however, the impression conveyed by the national polls is of a race in which Clinton bottomed out a few days ago perhaps after the FBI news last Friday and has now begun to recover.
But those state polls? Not a lot of good news for Clinton. Theres more data showing a tied race in New Hampshire. And Clintons lead in Pennsylvania is down to about 3 percentage points in our forecast. Polls in Michigan have also been tightening, with an unusually large number of undecided voters. Polling in New Mexico has been tight enough that were now considering it a state to watch, although that may reflect an abundance of caution. Clintons numbers have held up better in Wisconsin and Virginia, while the data has been very mixed in Colorado.
Just so you think Im not cherry-picking, heres every swing state poll weve added since our last Election Update at 7 p.m. last night. It includes the latest state-by-state tracking polls from SurveyMonkey, Ipsos and Pulse Opinion Research.
Read more: http://fivethirtyeight.com/features/election-update-national-polls-show-clintons-lead-stabilizing-state-polls-not-so-much/
"Just so you think Im not cherry-picking, heres every swing state poll weve added since our last Election Update at 7 p.m. last night. It includes the latest state-by-state tracking polls from SurveyMonkey, Ipsos and Pulse Opinion Research."
Even 538 worries about false accusations of "cherry picking".
Foggyhill
(1,060 posts)He's gotten that bad.
Too much punditry from this site now.
Helen Borg
(3,963 posts)"Just so you think Im not cherry-picking, heres every swing state poll weve added since our last Election Update at 7 p.m. last night. It includes the latest state-by-state tracking polls from SurveyMonkey, Ipsos and Pulse Opinion Research."
Foggyhill
(1,060 posts)The one he mentions to make a point.
Some of the polls he accepts should be rejected out of hand.
Not used and abused. He seemingly has zero understanding of polling compared to the 10 other poll aggregators.
Understanding stats (which he does seemingly) is not enough.
There is a hell of a lot of cooking the data without anything to really back it
(like some kind of science/math or experimental record in the field to back what he does)
He doesn't include early voting which makes his conclusions useless.
Helen Borg
(3,963 posts)Foggyhill
(1,060 posts)His methodology is NOT objective. Don't try that on me.
Please point me to how his cooking the data is based on actually tested methodology.
It is not. Not even close.
The only part he does that's OK is the stat part.
How he manipulates the data prior to that and analyzes the polls given them ratings, are skewing or whatever, is not.
If the data is not a random sample of the likely voter population., you got nothing about that population.
No amount of manipulation will give it to you.
Despite that, he doesn't use the one thing that actually gives you a pretty good sample of that, actual voters.
Helen Borg
(3,963 posts)Foggyhill
(1,060 posts)to razzle dazzle them
I hate people abusing math, science or tech in anyway.
If he was labeling his thing as experimental and then using his final results to adjust his overwrought and finicky
models and data gathering and analysis, I'd be OK with that. He could then maybe write a paper about it...
But that's not what he's doing. He is misrepresenting what he does.
As for the jealousy bit... Very weird comment.
I'm a women who has been 30 years in my very technical STEM fields, fields with few women initially;
I think I'm pretty much OK with myself...
uawchild
(2,208 posts)I thought he rated pollsters and weighted their polls by just doing statistics on how accurate their past predictions actually turned out, taking note of any bias trends in their results.
I am not a statistician or math wiz so I am asking you --- why is that a bad thing to do?
brush
(53,764 posts)Last edited Sat Nov 5, 2016, 01:06 PM - Edit history (1)
And many of them are Puerto Ricans who moved there because of the economic problems in PR, and they and other Latinoes won't be voting for Trump who called them rapists and criminals.
And that phenomenon won't just happen in Florida. Other states with large Latino populations are reflecting the same thing.
538 is not reflecting this so IMO they are no longer a reliable polling aggregator.
Trump will lose Florida, and thus the election.
uawchild
(2,208 posts)I always assumed that if a pollster called an early voter they would just ask who they voted for, just as they would ask someone who hashn't voted who do they intend to vote for. So, wouldn't that just be in all poll results anyways?
I have no idea though if early voters are less likely to participate in polls when called or feel less inclined to reveal who they actually voted for. That could effect polling, I don't know. 538 can only work on the polling data that is out there, so yeah, it could be a case of GIGO -- garbage in garbage out.
onetexan
(13,035 posts)"Democratic consultant Kevin Cate, who calculates the vote-differential between the candidates by averaging the polls into the ballots cast, said Trump is almost hopelessly behind. By Cates calculations, Trump is down about 90,000 raw votes.
'At this point, Trump would need to win over 50 percent of the remaining votes and that means its almost impossible to catch up. He needs an extra inning that doesnt exist,' Cate said."
I've stopped looking at 538 weeks ago. I don't think his poll is objective enough - it's more of a gut feel. The FLA Latino vote surge proves it. Hillary's got a slim lead but a lead nonetheless, and given such a high percentage trump needs to catch up, not to mention the unknown number of more Latino voters turning out in the general election on Tuesday I'd say Hillary's got the momentum in the sunshine state.
Coolest Ranger
(2,034 posts)paying attention to that sight but I just had someone accuse me of not living in reality and that said person only had about 208 post so i put him on full ignore
closeupready
(29,503 posts)if they just looked at the fine print showing his data.
Coolest Ranger
(2,034 posts)accept anything from five thirty eight. He can go jump off a cliff
Helen Borg
(3,963 posts)obamanut2012
(26,064 posts)I know this disappoints you.
kiri
(794 posts)I, too, am a physicist and I have taught statistics at univ level many times, and in advanced courses for PhD candidates. The criticism of 538 is unwarranted. This is exactly the same mistake that Romney and the alt-right made--"skewed polls". They believed the nonsense that since more self-identifying dems were included in some polls, the results were "skewed" and invalid. Romney's team so believed this that they went into shock when the polls turned out to have been correct; Anne went to crying. Many discussions/analyses show that self-reported party identification has a poor correlation with likely voting or candidate selection.
Critics of poll results are frequently victims of confirmation bias. This afflicts all of us, and its counterpart is to reject anything that we don't like to hear.
Nate Silver at 538.com has an excellent record of understanding the mathematics of statistics, comprehending the errors and complexities, and doing insightful analyses.
Any DUer who discounts 538 is not in the reality community. They are in the discredited skewed Romney community.
The fear is that Trump may win; 538 shows this a real risk. I am not glad to report this, but Trump may gain 270 electoral votes.
Coolest Ranger
(2,034 posts)statistics but i do care that you just insulted me by saying I don't live in reality.
jzodda
(2,124 posts)To not factor that into Nevada? Or Florida?
It can't be ignored.
still_one
(92,116 posts)for Hillary in swing states might I ask
I am
Chemisse
(30,807 posts)Every Clinton voter needs to feel their vote is urgently needed.
It could be that in the long run, Comey did us a favor by heightening the tension.
DCBob
(24,689 posts)Now that it turned out to be dud, as most of knew from the beginning, the numbers are returning back to where they were before...solid Hillary.
JCannon
(67 posts)...and I don't care: Over the course of several elections, the people most likely to slam Nate Silver are the ones who favor candidates in trouble.
Right now, BOTH sides are slamming Silver, which means that both candidates are in trouble. Hillary still has the advantage, and thank God for that. But she faces a real and serious challenge, and it is pointless to take a "kill the messenger" attitude toward the people at 538.
And that, of course, is your cue to call me a "concern troll" simply because I'm not a blinkered optimist.
Buzz Clik
(38,437 posts)Right, but most people have no idea how to read polls. 538 knows a lot (though their crystal ball was shattered in the primaries), and they benefit from an anxious electorate -- as do all media outlets.
Watch the campaigns over the weekend to see who's in trouble. The desperate campaign will be flurrying around in states that should be in the bag but aren't. The confident campaign will keep a steady beat -- shoring up support in solid states and trying to tip those on the bubble.
Buzz Clik
(38,437 posts)Michigan is not really tightening -- it is stable. 538's analysis shows this (seriously -- wtf!). New Mexico is wandering, Ohio seems to be a loss for Clinton. Pennsylvania is tight but holding, just as is Michigan.
538 proved two things during the primaries: a) their algorithms are horribly flawed and b) they didn't learn anything. Although this isn't the primaries, it's a different sort of election, and 538 won't admit they don't have any idea if they are on the right track.
538 has evolved in one important way from the past: they've become a viable commercial entity. They don't care who wins, but they really care about the traffic on their website. And, nervous voters will visit their site a helluva lot more than confident or despondent voters. So, let's promote chaos!
Two thumbs down on this one. Flush 538 and find a more reliable player with less money on the line.
obamanut2012
(26,064 posts)Buzz Clik
(38,437 posts)I'm going with the Princeton crew. They have nothing to gain from melodrama:
===========
And don't sweat Nevada. It's a gravy state. Trump needs a lot of Nevadas to make this competitive.
obamanut2012
(26,064 posts)I agree NV is a gravy state.
closeupready
(29,503 posts)broadcaster75201
(387 posts)nt
DeminPennswoods
(15,276 posts)To me that's the biggest flaw in his model. Pennsylvania is not like Ohio. The western part of the state is similar to Ohio and WVa, but the eastern part is similar to the Boston-DC corridor. You can't assume because Ohio or Iowa is trending one way that Pennsylvania is trending the same since the 2/3rd of the state's population lives in the eastern part of the state.
No matter how good the math behind the model is, if the assumptions are wrong, so's the model output.
andym
(5,443 posts)so if Hillary is recovering from the Comey story, then an increase in her national numbers will likely be reflected in the state numbers as well. The close state polls seemed to follow a few days after the Comey story and as that story recedes in people's memories, Hillary is gaining again.