This was my thought. When they said of course there were no shy Biden voters I rolled my eyes and stopped giving them credence. I remember hearing a very old story (from the UK) of a labour candidate knocking on a door to be told by the man of the house that it was and would always be a Tory household and then the woman sneaks out the side door and asks the candidate how to volunteer for them.
The idea there would be shy Trump voters but not shy Biden voters is just more of the rightwing fantasy that the rightwing is shunned and oppressed. Meanwhile black people (who overwhelmingly support Biden) are getting killed by the cops. Riiiight, it’s the rightwing people who have to worry about oppression.
The fewer undecided thing has really caught my attention. Polls still aren’t to be relied on, no matter what, but there is a really big difference between being “up four points” 47-43 and being “up four points” 51-47. Because in the first case, if the poll is 100% accurate you can still lose, not so much in the second. Biden’s numbers look at a lot more like the latter (again, Biden can still lose, but the comparison isn’t really there).
Finally, a comment about "getting it right". When we're talking about statistics and polling it is senseless to give credence to people who happened to predict the winner. You might as well call a person who hits on a 17 and gets a 4 a blackjack genius.
If your model gives a large group of candidates an 80% chance to win, you have a good model if 80% of them win, not if 100% of them win. It’s a good model if it’s right and wrong the right amount. People who tell you they can see the future are full of shit. I give some credence to Alan Lichtman because I think it’s actually possible to be good at making educated guesses (and have observed history professors in particular are pretty good at making guesses).
But for all the nonsense the national pollsters and Nate Silver’s model seems to have gotten the last election dead on. And these people’s looks like it was way off.
Hate to say it, and I doubt you meant to leave it out, but it’s not looking like a safe bet that crucial votes in various places will be counted (let alone the people who get polled but then basically aren’t allowed to vote).
Voter suppression on trump’s behalf seems sure to big much greater and more effective than it was in 2016.
I looked into exactly this and found that every serious polling outfit is giving extra weight to the non-college-educated-white demographic in Trump’s favor, given the lesson from 2016 was that they were the demographic most likely to secretly vote Trump. This notion that partisan quasi-laypeople can out-think the professionals is farcical. The true professionals know that if they get this one wrong, their entire industry is doomed for a generation or more. Now forget what I just said and vote like your life depends on it.
my secret wish is that the GOP attempts at disenfranchisement combined with their penchant for voting on election day will be their undoing this time… smart people (mostly dems) will have voted early or by mail, and the lines and hassle of getting to a polling place will leave some republicans angry enough to just not bother.
I am continually amazed that this race is as close as it is in so many battleground states. Why should someone like Donald Trump have ANY chance of winning a second term in those states? It just goes to show us how much of a force the right-wing hate machine is.
I’m hoping, praying, phone banking and drinking after my shifts this weekend. Short of forcibly detaining Trump voters so they can’t vote, I don’t know what else I can do.
Let’s just hope for a good outcome in the coming days. I cannot even believe we need to fight this hard for something so obvious.
It hits me every day that if Trump had done basically anything different re COVID he would be walking to victory at this point. All he needed to do Jan-Mar was check out and let the doctors deal with it.
Not BS. But it has caveats (and is born of caveats). The technique was taught in our graduate engineering classes in the 90’s. A summary of the lesson was “customers lie.” That you need to be extremely cautious about blindly trusting survey data. One example was GM asking people what kind of car they wanted (circa 1950’s or something). The data said affordable, fuel efficient, modest, etc. GM made the cars and they were flops. They started asking what their neighbors wanted for a car. Flashy, big, etc. GM made those and they sold great.
Other examples would be asking in 1900 what people wanted for transportation. A faster horse would be the winner. And how would you even know people wanted a VCR when they hadn’t even heard of one?
But what these guys seem to be missing is that if people don’t know their neighbor’s opinion, then they substitute their own, but if they think they know it, they will repeat it. Since there are some VERY vocal supporters all over the place, they may be overestimating support among neighbors and misleading pollsters in turn.
Just like our professor emphasized: be VERY careful with survey questions and data.
What happened in 2016 was not “shy Trump voters” or “the silent majority” or anything like that. It was undecideds (of whom there were still quite a few in the final weeks of the race) deciding, “Fuck it,” and going with Trump. A large part of that was James Comey. Clinton’s lead collapsed in the blink of an eye after Comey, and the polling that was done in the day or so before the election reflected what was happening, while older polling data (and many of what turned out to be swing states hadn’t had a poll in weeks) proved incorrect.
I’m so glad I read Boing Boing- I read this article elsewhere and started to worry. But I can count on you mutants to dig up and clearly share the straight dope … Thank you!
There’s no evidence that this method, however well it works for any manufactured goods, would apply to political questions. Because right now, despite a massive blue wave in 2018 and another one appearing on the horizon, every Democrat I know fears the Lazy Marmalade will get re-elected somehow, and we’re likely to shade our answers with that fear weighing down on us.
Frankly, I don’t fear if Bob next door buys Ford or Chevy, an economy car or a gas guzzler. I’m happy to take an educated guess based on what I know about Bob and his previous cars. I agree, one might answer the “good” answer but substitute our REAL desire when asked about our neighbor. It’s a car. It’s not the stake of the country, the existential fear that we are tumbling into an authoritarian oligarchy from which Democracy may never recover. And Bob next door is probably as likely as I am to vote Biden, but I’d fucking guess Trump because I’m that fucking scared for the future of my country.
In short, this method is hugely flawed, and I’m banking on that to help keep the fear at bay right now. I may be deluding myself, but it’s what I got and I’m sticking with it.
