|
"Vote Hillary Clinton 2016" is dead. Long live "Vote Hillary Clinton 2016"! (pg. 7)
|
View this Thread in Original format
| Vector A |
It's not just the HuffPost, though. It's NYT, Washington Post, and also independent academic political analysts that are giving her significantly better odds than Silver is, from what I gather.
Still, he was bang on last time, so I give high weight to his assessment. |
|
|
| DJ RANN |
| quote: | Originally posted by Vector A
It's not just the HuffPost, though. It's NYT, Washington Post, and also independent academic political analysts that are giving her significantly better odds than Silver is, from what I gather.
Still, he was bang on last time, so I give high weight to his assessment. |
That's exactly what I'm getting at. Much as the huff is biased to the left / hrc, it's still a little out of the norm to go right at nate silver like this, and they aren't he only ones saying the vote has been "underestimated" by 538. I'm also cynical enough to believe that for the right price, just about anyone in media, let alone polling stats will say whatever you want them to say, so bending some incredibly complex polling data isn't going to permenetly damage anyone's brand, especially when there's plausible deniability due to that complexity.
Completely anecdotal, but my neighbor was the cheif editor of one of the big British tabloids for over a decade and now works in TV here.
He has absolutely no doubt the media is serving it's finincial interests by painting this as a close race. He stopped short of saying he did that all day long.
But I've always trusted 538 so I'll take their word until proven otherwise. |
|
|
| Jon_Snow |
| quote: | Originally posted by Vivid Boy
Tomorrow grab freedom by the |
This election is up for grabs. |
|
|
| Woony |
The 538 thing is quite simple. They underestimated Trump in the primaries and so they adjusted their model to assume a larger margin of error towards the runner up than other pollsters have. 538's percentage is INCLUDING the possibility a systematic polling failure.
Other pollsters like Princeton have Hillary at like 98% because if you look at the state polls and don't assume a DRAMATIC general polling error of over 3% (which I don't believe given the data we have on early voting and registrations), it's basically impossible for her to lose. Brexit comparisons don't make sense here because Brexit was closer and polling for the US election is basically the most advanced polling there is, Brexit polling was amateur hour in comparison.
And of course, 538 is a for profit site. I don't think they are falsifying data or anything but they choose models which are more "entertaining", whereas non-profit pollsters like Princeton choose more "boring", conservative models.
Btw, you know what's great? If Hillary wins by around 4-7% it proves what some smart people have already suspected: the whole media circus, the polls swinging back and forth every day, it's all just for show. Nobody actually changed their mind from what they thought in Augut. The now polls average around +4% for her and that was also the number her internal polls apparently have had since summer. Add some points for superior GOTV and there you go. All that actually fluctuated is the rate at which Hillary/Trump supporters respond to polls depending on what's in the news. I mean it makes sense, the idea that 10% of the population would go back forth between these two candidates is ludicrous. |
|
|
| Jon_Snow |
| quote: | Originally posted by Woony
nobody actually changed their mind from what they thought in Augut. |
:conf: |
|
|
| Woony |
| If someone said in August that they are voting Trump, they were going to vote Trump the whole time and are going to actually vote Trump now and vice versa with Clinton. The polls made it look like there was this massive fluctuation between the two which (probably) didn't actually exist. |
|
|
| Jon_Snow |
| This is most likely the case in all elections. Very few people who vote deviate from party lines. |
|
|
| SYSTEM-J |
| quote: | Originally posted by Woony
If someone said in August that they are voting Trump, they were going to vote Trump the whole time and are going to actually vote Trump now and vice versa with Clinton. The polls made it look like there was this massive fluctuation between the two which (probably) didn't actually exist. |
That's setting aside the pertinent fact that these are two remarkably unpopular candidates. It's perhaps less that people are switching between Trump and Clinton and more that people who are undecided or who are thinking about keeping their hands clean by voting for a third party could be horrified into changing their opinion by some of multitudinous muck that has been raked over the last few months.
The defining message of the whole campaign has been "The lesser of two evils". Young people, normally the most liberal demographic, aren't behind Clinton. You can easily see these more right-on voters might sway away from Clinton every time there's an email scandal, or back towards the necessity of voting for her when Trump is painted as particularly egregious. |
|
|
| Woony |
Third party support isn't big enough to account for the swings in the polls. I don't think there ever have been a lot of undecided people this election, for most one evil is clearly much bigger than the other. Millenials may not love Clinton but they are horrified of Trump. Old conservatives may not like Trump's crass behaviour but they've gone through 30 years of anti-Clinton propaganda. Voting Turnout is already projected to be higher than in previous years in many locations, if people really had difficulties to decide, the turnout would be rock bottom with two unliked candidates.
And the thing is, if a young millenial or some kind of redneck blue-collar worker that has never voted before decided to stay home instead of voting, that doesn't show up anywhere so it's hard to measure. |
|
|
| SYSTEM-J |
| Your logic essentially seems to write off the existence of floating voters. |
|
|
| DJ RANN |
| quote: | Originally posted by Woony
The 538 thing is quite simple. They underestimated Trump in the primaries and so they adjusted their model to assume a larger margin of error towards the runner up than other pollsters have. 538's percentage is INCLUDING the possibility a systematic polling failure.
Other pollsters like Princeton have Hillary at like 98% because if you look at the state polls and don't assume a DRAMATIC general polling error of over 3% (which I don't believe given the data we have on early voting and registrations), it's basically impossible for her to lose. Brexit comparisons don't make sense here because Brexit was closer and polling for the US election is basically the most advanced polling there is, Brexit polling was amateur hour in comparison.
And of course, 538 is a for profit site. I don't think they are falsifying data or anything but they choose models which are more "entertaining", whereas non-profit pollsters like Princeton choose more "boring", conservative models.
Btw, you know what's great? If Hillary wins by around 4-7% it proves what some smart people have already suspected: the whole media circus, the polls swinging back and forth every day, it's all just for show. Nobody actually changed their mind from what they thought in Augut. The now polls average around +4% for her and that was also the number her internal polls apparently have had since summer. Add some points for superior GOTV and there you go. All that actually fluctuated is the rate at which Hillary/Trump supporters respond to polls depending on what's in the news. I mean it makes sense, the idea that 10% of the population would go back forth between these two candidates is ludicrous. |
But this is exactly the point; 538 are part of the "circus" and in some ways they are both feeding the problem and feeding from it - Study after study has shown (as is human nature) that those that are ahead do better and those losing ground tend to lose more ground. The compression a rarefaction of support for either candidate is really nothing more than using hype as a pulse, a machine reciprocating it's own movement.
Not withstanding jack's correct sentiments about the lesser of evils, and the Berniebro factor, there really weren't many "undecideds" and honestly if you're really that ing undecided after having politics rammed down your throat for the last 18 months as we have in the states, then you probably shouldn't vote anyway. We've never had so much information about candidates, especially from candidates that are not just people but long established brands, and have been in the public eye for over 40 years each.
If you don't know by now, then you never will and that's why i think all this wild sway of numbers is complete fantasy.
There may be some people who weren't going to vote than came out for fear of the other candidate but i don't think that sways the numbers one way or the other in any meaningful amount.
The interesting thing is that 538 have wildly changed their numbers over the last 24 hours; Clinton went from being at 64% chance of winning to suddenly 72% as of right now (and that's before any results have come in). Sure the final polls came in yesterday but and 8% 24 hour swing? It really may not be 538's fault but really, they're mean to be an aggregator, sorting through the mess and all it proves is that polling can't be trusted even when you aggressively attempt and use incredibly complex models to make sense of the numbers. |
|
|
| Woony |
| Multiple states are a toss up, so even slight changes to the model can have big effects. 538's model is just a result of people that want "live 24/7 polls" which doesn't make much sense honestly. But that doesn't mean that polls can't be trusted, they predict the winner without fail (at least in US presidential elections) and usually are only off a couple of percent. Polls are good at showing a median which isn't too far off from the result, they're not good at predicting states where it comes down to a few thousand voters. |
|
|
|
|
