@Ogion The Indy star is also a very bias source and wasn't listed in the sources I checked. The rest were toss up or leaning McCain. All I'm trying to get across is that the polls aren't always accurate.

On the subject of a racist GOP I don't know what GOP debates you watched but the only rhetoric ever mentioned on the subject of race was that of helping prevent illegal immigration which costs billion upon billions of dollars for the United States. Trump is the only one that is even close to be racist and even that is just liberal media trying to distort what he means from what he said early on in his campaign (which I still firmly believe he started as a joke).

@TrPrado Well no candidate has won any state so how can they possibly predict win probability from that? If they use primary election votes then republicans should be winning in almost every state as they usually have a larger primary turnout than that of typically blue states.

People give weight to 538 because it is bases sea on evidence and statistical analysis not random bbullshit. It is quite scientific in its approach. However people never seem to realize that there is no real way to test whether it had the probability correct. If candidate a wins it is hard to know whether it was a rare outcome or an expected one. Instead, 538 is a statement about where the preponderance of evidence lies. pEC might be methodologically superior however

They use the probability of who will win individual states. These people have studied politics for years, they know quite well that states matter.

Oh yeah. That's why republicans take the issue of police brutality so seriously and never try to deprive people of color of their votes. I am old enough to remember Wilkie Horton and the decades of racist dogwhistling going back to Reagan.

@Ogion I do respect their analysis but even they admitted in October that their data disagrees and that this is a very hard election to predict. Even today they have articles saying it's a very close election and that they can easily be off.

/\ liberal lapdog

@Pompeii You're profoundly ignorant about how politics works.

Also statistics. They're not saying that they can easily be off, they are saying that there is a 25% chance that Trump wins. That is not saying that their model is wrong.

Your name is hilariously accurate

It gives me the freedom to be my true self.

They have been weighting uncertainty more highly than other outfits, which is the o key reason they have Trump at 25% not somewhere between 3 and 15% like everyone else. 538 has the uncertainty baked in

@Condenscension I'm not ignorant I'm just hopeful. The Field Research Poll which has (according to 538) a 100% accuracy in predicting elections has stated a trump victory isn't improbable and it's going to be a very close race that they think he'll edge out in. I'm still basing my arguments on factual data.

They recognize there is a chance for him to win, but they doubt he will.

I've seen 538's polls jump around too much this election for me to view it as the *best* source. I've stuck with FRC and FieldINC and based on what they had it contradicts what 538 has had despite 538 admitting they both have a 100% accuracy in election prediction.

Abge is a single issue voter. (Marriage)

"Jumping around" is not evidence that a statistical model is bad. It is evidence that it takes new data into account.

Not when it shifts support of one candidate dramatically to another in the matter of a couple hours multiple times.

Condescension needs to be put up for best name of 2016

Were you looking at nowcast? That has the tendency to spike dramatically, but polls plus has far less dramatic trending.

No it was 538's. It's not too irregular but still made me check out a few other sources.

Condescension needs to be asked to provide an argument/rebuttal with substance.

538 has had 3 ways of showing the election. Nowcast being sporadic and not invented to actually predict results, and polls plus being expected to be their most accurate prediction.

Indeed, bouncing around is what it should do if the data is noisy. If you have a criticism of 538, please explain what you see as wrong with their methodology. For my part I think the main issue is the frequentist approach to incorporating new info. Of course the only Bayesian model I know of has Clinton at something like 99% which intuitively seems difficult to see. However, we do have a megashitton of data that trough lay all points in the same direction, so perhaps not

God looking at the crowds at Trump rallies is chilling. Reminds me of the Nuremberg rallies. I'm still waiting to hear them burst out in chants of Sieg Heil

I personally believe that betting markets have the best predictive value due to their holistic nature, but that's a hard thing to verify. Polls-plus and betting markets have been pretty close so far.

I don't recall there being any bounciness in the polls-plus model. @Pompeii, you may have confused yourself by seeing stats from different 538 models and assuming that they were rapidly changing the predicted outcome. Earlier in the election there was very little polling even in the swing states, so several polls coming out on the same day in multiple swing states can cause swings in the prediction.

Towards the end of the election, it also makes sense to swing heavily because those polls are more likely to represent the endgame distribution.

You'd have a hard time finding any statistician that derides 538's model for "swinginess". There are other criticisms, but that isn't one.

I'm only deriding it because the nature of it's shifts are somewhat contradictory to other models from my other sources such as FRC, F&A, MassINC, SurveyUSA, and Marist to name a few. I fully understand polls will change due to early voter turnout and registered voter numbers, but 538's changes in contradiction with other sources they recommend has caused me not to use it as my primary source. I haven't said their model is completely inaccurate as you seem to think I've claimed. I simply believe Trump has a better probability of taking swing states than 538 gives him credit for. Many polls coming in now seem to contradict their predictions slightly, but the day is still very young.

Everyone get read for cars burning in the streets, people being bludgeoned to death, savage sodomizings, ritual sacrifice of animals; trumps gon' lose.

The most hate filled violent political movement in 1000 years is about to get defeated. Imagine how angry and rabid these animals will be.