Earlier today, Huffington Post’s Senior Polling Editor Natalie Jackson ended a rambling diatribe on why polls showing Donald Trump ahead should be ignored with the following conclusion:
The numbers themselves aren’t reason to label a poll an “outlier” ― sometimes the first poll to show a trend unfairly gets that label. However, there’s no contextual reason to believe that there’s a trend of support moving toward Trump. The campaign is flailing, and there haven’t been any new Clinton scandals of the magnitude it would require to reverse her sizable lead.
Bad samples or other errors happen to even the best pollsters. When the Trump campaign or supporters try to leverage the three national polls that show him leading, keep in mind that more than 30 in October have shown Clinton ahead. The likelihood of that many polls being wrong is very low.
Let’s ignore the obvious propaganda RE: Clinton not having any destabilizing scandals: “Wikileaks”, “Project Veritas”, and “Clintons in Haiti” are easily searchable alt news terms.
Let’s even ignore the idea that a groundswell of organic Trump support doesn’t exist: a glance at the live stream viewership of Trump’s daily rallies on channels like Right Side Broadcasting might convince you otherwise.
Instead, let’s focus on the polls.
An actual statistician – not a “Senior Polling Editor” – recently explained why Clinton’s apparent lead in so many polls means nothing if the polls themselves are inaccurate. His work analyzing data from 7 popular polls was featured on ZeroHedge earlier this week, and includes the following startling analysis:
The average margin of error on these 7 spreads shown is only 3%. Most polls should therefore be within a few percent of the 6% average spread that is advertised by media. But instead most are not!
For example, the difference between the highest Ms. Clinton spread and the lowest Ms. Clinton spread is >14 percentage points! And the standard deviation among these mainstream polls is 5%. So both have to be added together, and each is already higher than 3%! That’s an unusual, impossible outcome through luck alone. Therefore something is misrepresented in the polls.
Also right now 2 of the 7 polls favor Donald (you just’ don’t hear about them), so double the 10-15% odds he is being given. In the final analysis of this trinomial data, on November 9 we’ll look back and see only one poll being correct and most were flat out wrong. This evidence below is a breach of the probability theory behind proper polling, where most polls should see the correct spread within the margin of error interval (that’s what the interval’s definition must be!) If the margins are therefore completely busted, then so too are the egregious spreads that are seen to be all over the place (and mostly untrustworthy). Likely the correct expected spread right now is 4-5%, and the larger spreads are coming from pollsters that ironically also have the highest margin of errors (casting further suspicion on how close the election really is for Americans). We stand by our long-running estimate that the current probability for a Donald Trump victory is about in the 20% range, or twice what mainstream media is projecting. Of course that is low, but to some it’s still a compelling 1 in 4 chance (and much different than some might expect given all the twists and turns this campaign season has brought us). It’s also a better reflection of the true odds, versus those dished out by the same inane talking heads who recently gave you the Brexit “remain” prediction, or the NeverTrump prediction!
Speaking of Brexit, monetary bets on the outcome of the presidential election are beginning to adapt a pattern similar to the trends seen leading up to the EU referendum. Although 71% of the money is behind a Clinton victory, 65% of the bets are for Trump, and English bookmaker William Hill has boosted Trump’s chances of winning to 20% as a result. Check out some expert analysis in this Independent article and decide for yourself:
William Hill’s spokesman and resident betting expert Graham Sharpe, an industry veteran of 44 years standing, said: “It’s very, very similar to the Brexit vote. There is a metropolitan media bias that says Trump can’t win, but they can’t vote. In betting terms, this is not a done deal. I see parallels with the Brexit vote at this stage.”
Mr Sharpe, the author of more than 25 books, mostly on the betting and racing industries, continued: “Some of the larger number of small bets on Trump can be explained by odds. When you have odds of 1-8 or 1-9 as you do on Clinton it’s not so interesting to the small punter. You tend to get bigger bets. We had one woman walk into a Northumbrian betting shop to stake £170,000 at 1-8 to win £21,250. She came in the next morning to stake another £13,200 to win £1467. She’d never placed a bet before but said she saw it as an investment.
“Trump at 4-1 or 11-2 will attract the smaller punters. But I don’t think that is only reason for the number of bets on him, and we have had to cut the odds on Trump three times in the last couple of days. This isn’t over.”