Pollsters Warn of Flawed Election Day Predictions: ‘People Are Going to Be Shocked’ by ‘Shy,’ ‘Hidden’ Trump Voters

Notable pollsters are cautioning Americans to refrain from getting too comfortable with polling predictions as Election Day approaches, suggesting polls may not account for “shy” or “hidden” Trump voters– those who are largely responsible for President Trump’s upset victory four years ago. In this election, the magnitude of such voters could be even greater.

All eyes are on key swing states, such as Pennsylvania, Wisconsin, Michigan, Florida, North Carolina, Arizona, and even Minnesota as Election Day swiftly approaches. Over the weeks, Breitbart News has spoken to dozens of Trump supporters across those states, Florida and Wisconsin, particularly, who also expressed the belief that the silent majority will show up in a greater volume than in 2016.

“We have everything, and I said, well, after hearing that they were going to pack the court, I finally said I’m putting my Trump flag up,” Bryce, a Wisconsin resident, recently told Breitbart News.

“Wouldn’t you know? Dozens in my neighborhood went up. The lesbian couple across the street — they’re very nice people; we love them — I didn’t think they’d ever put one up. They’ve got a Trump flag. There’s a lot of support that is sleeper support,” he continued.

“It’s like you gotta just ‘shh’ — just be cool,” he said. “Show up to the polls and vote, you know.”

“It’s like it’s a secret but when you meet somebody else in the secret society that supports Trump — because you can’t do it publicly or people will rip you. You know,” Bryce added when asked if there is a bigger and stronger silent majority from 2016:

Indeed, you cannot. I analogize it to the example of a new drug trial. Say you enroll 10,000 people in a drug trial to see if the drug works. But 9,800 of the participants drop out of the trial for reasons you don’t know. You are left with 200 people who took the drug and of those you got 46% who saw improvement and 45% who saw no improvement and 9% who can’t tell if the drug helped them or not. This trial would be thrown into the waste bin by any measure of any reputable scientific analysis and certainly would never be approved by the FDA. And yet it seems we let the mass media sell us this “junk science”.

And you also hit on another point, which is how to distribute the answers when 60% of the 200 people are female age 24 or under. Well, what one would try to do is censor the data according to demographics. If the voting public at large is only 5% women under 24, then you would chuck out 95% of the responses and then try to extrapolate the responses by the 40% who may be more representative (men, people aged 25-44, people 44-65, people 65 and older etc) but you would only have very limited number of real people in each of those categories. So maybe you got 2 here and 3 there, and you try to extrapolate that tiny sample onto millions of people? It’s complete rubbish. You actually have no clue - 1 of the 3 may be an outlier (e.g. a complete lunatic who thinks Elvis is still alive playing music with Bob Marley on a Greek island) so you extrapolate that 33% of that demographic are complete lunatics. And still, even if it were correct, you still have no guarantee what the actual voter turnout makeup will look like. Maybe it snows. Maybe there’s a hurricane. Maybe they just prefer one candidate but really aren’t all that motivated to stand in line to vote. It just doesn’t work.

And it is no coincidence that all the polling outfits show a “tightening race”. They publish what the people paying them want them to publish by tinkering with the models, but as election day approaches they tinker some more in order to preserve their reputation so they can get hired again next cycle, telling everyone “we were pretty close, and none of the other pollsters got it any differently.” It is a business, after all. None of them want to be so far out of the pack because if they are wrong they won’t get many gigs next year. So they all come in with basically the same numbers on purpose in the last weeks of the election so none of them are singled out as being way off when they ask for jobs later.

Thanks for bringing some clarity to the concepts I was trying to articulate.

Based on your points and my own understanding, it seems like phone surveys are so obsolete and out of touch with 21st century reality, their results have NO and I mean NO basis to say they are forecasting what will happen on election day.

If your 800 poll participants, instead of answering questions on the phone, instead threw darts at a dart board, the forecasting results, it seems, would be no less accurate.

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.