Posted on 11/01/2020 3:23:02 PM PST by Widget Jr
Ciara Haley and George Szamuely talk to renowned litigator and political analyst Robert Barnes about polls and why they are of such limited usefulness.
Problems with current polling:
1. People are pretty bad at predicting how they will vote in the future, don't know, or lie to pollsters.
2. Sampling and representation have gotten worse. The bigger the class divide, the harder it is to poll. In 2010, people stopped responding to polls. Political polling used to have a 50% response rate. It is now down to 2%. Polling is no longer random or representative.
3. Pollsters know the current polls are wrong. The polls the press are talking about predicted the early vote for all fifty states. The error in early state voting has an average error of 25 points.
On top of that, the enthusiasm indicators all point to Trump.
Nope. They’re inevitably wrong.
No.
No.
Getting near “last dance” time for Biden... ;-)
Yes, #2 is especially troublesome. It is called selection bias. The pollsters only get responses from people who want to talk to pollsters. These are either the most energetic supporters, the most lonely for the attention, or completely insane people. Not at all representative.
In any case, a legitimate polling outfit would have to throw out a huge percentage of the responses to adjust them and extrapolate them out among the various demographic groups that the pollsters think will turn out, which they really couldn’t know as their projection is based on historic turnout which isn’t necessarily indicative of the future especially in times like these.
Maybe Kamala should try last ditch pole dancing for votes.
The exact same kind of anti-Trump attitude that drives the lamestream media (NY Times, LA Times, New Yorker, Time, NBC, ABC, CBS, CNN, MSNBC, etc.) drives the left-wing pollsters.
Poll takers or poll aggregators--makes no difference. 90% of the polls are very heavily slanted against President Trump in order to dishearten voters and negatively affect the Trump vote.
In short, the great majority of polling for this election cycle is totally worthless.
The exact same kind of anti-Trump attitude that drives the lamestream media (NY Times, LA Times, New Yorker, Time, NBC, ABC, CBS, CNN, MSNBC, etc.) drives the left-wing pollsters.
Poll takers or poll aggregators--makes no difference. 90% of the polls are very heavily slanted against President Trump in order to dishearten voters and negatively affect the Trump vote.
In short, the great majority of polling for this election cycle is totally worthless.
Demographics is not even good enough.
To get an accurate poll (if anybody actually cares about such a thing.. :-) ), you need to _validate_ the sample.
That means the sample must match known data.
So, for example, if the voters split 50-50 in 2016 in a state (Trump-Hillary), your sample _must_ have split 50-50 Trump-Hillary which means you _must_ ask that question!
I don’t believe any of these pollsters have done that validation.
They are fundraisers. Nothing more.
Democrats have faith in polls like they do in government...
You can pay attention or not, neither has any effect as long as you vote.
It’s a good and fair point from a statistical perspective. But it’s still not going to be predictive. Different times and issues, different candidate(s), and to some degree different voters. This year in particular with civil unrest and lockdowns and the virus and unemployment, it’s gotta be extremely hard to predict what will motivate people and who the voters will blame for the issues that impact them the most.
I do think, but it’s just my opinion, that Biden voters may be the most vocal (at least until the last few weeks) but also the least inspired to vote. They would be more anti-Trump than pro-Biden and I like to think people want something to vote for, not against.
You will probably will enjoy this—one very professional pollster said that many people don’t actually know how they will vote once they get into the voting booth.
He gave the example of how we always tell ourselves and others we are going to do something next week, and then for whatever reason, we just do something different.
So—I agree that even the very best poll cannot avoid being off a fair amount now and then.
We shouldn’t, but it is human nature that we will and the biased MSM knows it.
Scott Adams did a non-scientific poll: have you been polled, and were you honest when you answered? A large number of those polled said no, I didn’t tell them I was voting for Trump.
I think the best way would be on the street, in stores, waiting on car repairs, ect. Not sure why they haven’t started that. I think it’s ve more accurate then the phone.
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.