Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie
US Presidential Election 2020
Options
Comments
-
-
-
-
I feel a little bit dirty Pete agreeing with you on this
Well, I called the Russia-Collusion accusations to amount to be the nonsense it undoubtedly was, the Impeachment nothing burger to come to just that, nothing, and that the Steele Dossier was in fact used in the FISA court when it was claimed that it wasn't. I'll soon be proven right on Flynn and also on Trump's reelection. So don't feel dirty agreeing with me, Tippex. Feel assured100% something should be done about it. The crazy thing is that it seems to be both sides are in on this. I always said politics is a dirty game.
Well, the woman in question released a statement and is claiming that she knew they were up to no good and went along with it so she could expose them. Interesting defense. Guess we'll soon find out if she's telling the truth or not.0 -
-
Advertisement
-
-
-
-
-
-
Advertisement
-
-
-
-
-
The data just doesn't support your hypothesis.
Taking a quick look at the results of FiveThirtyEight's 2018 Midterm Forecasts where the results of 506 races were forecast, here's how they did against the final results:
All Races Lite Forecast 95% Correct (92% expected)
All Races Classic Forecast 96% Correct (93% expected)
All Races Deluxe Forecast 97% Correct (expected 94%)
If you have data that supports the shy-voter/pollee theory, it would be great if you could link to it. Without it, it would be just as easy to suggest there are significant numbers of registered Republicans who are actually shy-Biden voters.
The very nature of a "shy voter" is that they are highly resistant to being captured in your traditional metrics. They don't wish to engage with your analysis, and so are for all intents and purposes invisible in many statistical models.
If roughly 98% of people are not engaging with something like opinion polls, then you have a potentially giant blind spot. And you can't really accurately put these people into definitive categories, other than being "non-respondent/non-compliant" individuals.
If you had a blind spot that big while driving a truck or bus, you would be running over a huge amount of people without ever knowing they were even there!
Many Trump voters are basically ghosts... you don't know they exist, until they cast their vote. And then many will just go back to being ghosts again afterwards. The exit polls in 2016 seemed to show this too, as they were predicting a very strong Clinton win at certain points. So it wasn't just the polling prior to the vote that got it wrong... even afterwards they were still getting highly inaccurate forecasts.0 -
-
-
Gretas Gonna Get Ya! wrote: »The very nature of a "shy voter" is that they are highly resistant to being captured in your traditional metrics. They don't wish to engage with your analysis, and so are for all intents and purposes invisible in many statistical models.
If roughly 98% of people are not engaging with something like opinion polls, then you have a potentially giant blind spot. And you can't really accurately put these people into definitive categories, other than being "non-respondent/non-compliant" individuals.
If you had a blind spot that big while driving a truck or bus, you would be running over a huge amount of people without ever knowing they were even there!
Many Trump voters are basically ghosts... you don't know they exist, until they cast their vote. And then many will just go back to being ghosts again afterwards. The exit polls in 2016 seemed to show this too, as they were predicting a very strong Clinton win at certain points. So it wasn't just the polling prior to the vote that got it wrong... even afterwards they were still getting highly inaccurate forecasts.
Other than thoughts and feelings, have you any documentation to back this 'shy voter' theory up or are you just parroting buzzwords0 -
-
-
Advertisement
-
-
everlast75 wrote: »Pollsters were quite accurate in the 2016 election, within the margin of error.
Do you think they have not improved or refined their analysis in the last 4 years? Surely they have an incentive to do so considering it's their reputation on the line.
Clinton had an 87% of winning the election a week out, and 71% chance of winning the day before according to the polls... so basically the overwhelming favourite to win. Which was hugely inaccurate.
With 4 years of publicly bashing Trump and anyone who dares to back him, it's very natural to expect that many of these people will go even deeper underground with their views/opinions than 4 years ago.
Also, something simple that is very often overlooked... the nationwide polls are fairly useless predictors of the outcome of an election. Even if accurate, they predict the popular vote. Blue dominant areas of the country, can and very likely will become even more entrenched in their views... so this adds millions of extra support to Biden, without necessarily giving him any greater % chance of victory overall...
In the red states, and even many of the swing states, there is huge anecdotal evidence on the ground that support for Trump has actually increased rather than decreasing. And this anecdotal evidence, which was largely dismissed 4 years ago, very often proved to be a more accurate predictor of the outcome than the polls.
But of course this irks many people, because it's not particularly scientific... but not everything in life is scientific or easily quantified.
Even just look at the opinion poll in the CA forum... overwhelmingly in favour of a Trump victory. Despite the fact that the comment section of that thread is dominated by very vocal anti-trump rhetoric. Of course this will be casually dismissed by many as just being down to re-regs... but there is no way of proving this definitively... it is very plausible that there is a silent majority backing Trump even on this site (Not that everyone backing a trump win is necessarily pro-trump of course - but many could be)... However, the silent minority are very vocal, because it is more socially acceptable to publicly take the anti-trump stance.
Even myself... I'm very vocally pro-trump on this site. But other than among my family and closest friends, I do not give my opinions freely out in public. Because I'm smart enough to know that it might bring some unwanted and wholly unnecessary criticism. Most people simply do not want that hassle in their daily lives.
If 98% of people are "poll shy"... that's a potentially huge blind spot. If that many people can be poll shy, then plenty of people can potentially be shy in many other ways. Again, very plausible.
I'm sure the pollsters will be trying very hard to refine their approach... but I wouldn't be staking much on their ability to get it right in this election cycle. They may get a handle on these populist type movements eventually... but not in the immediate future. I think we are going to see huge discrepancies between the mainstream polls, and the outcome of this election. (That's if they don't deliberately tighten them before the final count in order to save their reputation)
But it will hopefully open up discussion on many aspects of how we vote and how we form opinions in society... the polling industry itself, is less of a concern to me personally. I think it's shining an important light on certain aspects of our society... but whether or not traditional polling survives... frankly, it's a bit of a side issue. (imho)
0 -
-
-
-
Gretas Gonna Get Ya! wrote: »Clinton had an 87% of winning the election a week out, and 71% chance of winning the day before according to the polls... so basically the overwhelming favourite to win. Which was hugely inaccurate.
This is incorrect and also misleading — on November 1st 2016, Clinton had a 71.2% chance of winning according to 538. On November 6th 2016, she was at 64.9%. On election day, November 8th, that ticked back up to 71.4%, but polls in the last couple of days are very susceptible to herding, so it's never wise to look at any that are that close to the election.
Beyond that, you're conflating what the polls say with what forecast models calculated based on those polls. Let's look at the actual polls. Here's the set of national results:
Noting that the margin of error on national polls is around 3-5 points, and the fact that Clinton indeed won the popular vote by 2 points, none of these results are outside the margin of error!
What was a more critical problem was the poor weighting of voter education caused errors in polling in PA (Forecast Clinton +3.2, Result Trump +0.7), MI (Forecast Clinton +4.2, Result Trump +0.2), and WI (Forecast Clinton +5.3, Result Trump +0.77). These are the states that swung the election for Trump. Only the Wisconsin forecast is outside the margin of error.
There's a sound reason why margins of error exist, and why they're so important for people to understand when they're looking at polls and forecasts. But the polling errors didn't come from shy voters, they came from the pollsters not weighting by education — something most of them have corrected this time around.
This is also why polling averages and forecast models are so much more useful than individual polls.0 -
-
-
-
Advertisement
-
This discussion has been closed.
Advertisement