What's Wrong with the Polls?

[LanguageLover]

Last edited Mon May 8, 2023, 02:28 PM - Edit history (2)

The top three Biden Popularity Polls today on 538 show the following:

Rasmussen 51% B rated pollster
TIPP Insights 43% A- rated pollster
ABC News 36% A rated pollster

Firstly, how can there be a 15% difference between Rasmussen and ABC? Rasmussen is not a Democratic leaning poll by any means. A 15% difference between two polls taken at more or less the same time? It is unthinkable.

Then, sitting right in the middle is TIPP at 43%. Just because this looks to be an "average" of the other two doesn't mean that it is correct. It could very well be wrong too.

Something is very wrong with the polls. I am not one to say that I never believed in the polls because I do. I regularly look at 538 and Biden's popularity numbers. However, there is something wrong in the very methodology of "collecting the data" that different pollsters are using, to cause this array of numbers which continue to appear daily with none having any correlation with another.

It has been going on for quite some time. Let's look back at some of the Poll Prediction Averages which were put up by 538 before the Nov-2022 midterms:

NH Maggie Hassan was predicted to win by 4% but finally won by 10% (diff +6)
PA Fetterman was supposed to LOSE by 1% but finally won by 5% (diff +6)
WI Barnes was supposed to LOSE by 5% but finally LOST by 1% (diff +4)
GA Warnock was supposed to LOSE by 1% but won by 2% (diff +3)
CO Bennet was supposed to win by 8% but won by 15% (diff +7)

538 gave the Democratic Party only a 41% chance of winning the Senate. I am assuming that to mean that they gave the Dems a 41% chance to get to just a 50-50 Senate, which would technically be "victory" for them. The Dems cleared that easily and actually got 51-Seats finally.

And here are two Governor Races in the same election:

PA Shapiro was supposed to win by 10% but won by 15% (diff +5)
WI Evers was supposed to LOSE by 0.4% but won by 3.4% (diff +4)

I have tabulated the above just to show that in all of the above the Dems beat 538's average predictions - and quite comfortably too in the majority of the cases. Only in Wanock's case was the difference +3 over prediction but in the others they were 4 to 7 percentage points higher.

There is something very wrong with the polls folks. I have a feeling that the "demographic representation" while collecting the raw data of opinions, has become very faulty for whatever reason and therefore we are getting numbers which no longer represent reality.

The problem for me and others like me is that we don't know whom to believe. Any Dem should be concerned about the ABC poll showing Biden at only 36% right now but then all he has to do is to look at Rasmussen showing 51% and immediately be comforted again. No one knows what's going on though.

To end this discussion, a last one: Before the Wisconsin Supreme Court election a month ago, a Dem operative said on TV that he would be happy to win it by just 10 votes! That's +10 votes total out of hundreds of thousands cast! The Dem backed candidate won it by a whopping 203,000 votes.

[hippywife]

1. First off...

I never pay attention to polls; they're not terribly accurate and can be unnecessarily anxiety inducing.

Second, why does anyone still trust anything from 538, even with Nate Silver supposedly gone?

[milestogo]

15. Thank you.

[spanone]

2. ...the only one cited all morning has been the ABC poll....go figure

[Mad_Machine76]

8. It's like a dog chewing on a bone

The media just won't let it go. They just want to hammer Biden even over just one poll that might be an outlier for all we know. But there is nothing like bad news for Democrats for the media to endlessly promote.

[jimfields33]

10. Polls are used for if the election is today.

[LakeArenal]

3. I don't believe in polls. That's my opinion.

My opinion is they are manipulated.

My opinion is they are antiquated.

My opinion is they are as accurate as weather reporting. Meaning I have as much success just by looking out the window. Or, in the case of politics, what I hear on the streets.

A year ago… 538 king of polls. But that was a facade.

[NoRethugFriends]

4. Actually, weather reports are pretty darn accurate these days

[leftyladyfrommo]

21. Not in KCMO. we get horrible storm warnings all the

time that just evaporate once the storms hit Tonganoxie. They go south of us or off to the NE. Thank goodness because the storms have been pretty violent lately.

[onecaliberal]

5. ABC loves propaganda.

[yardwork]

6. Who answers the phone now?

The ABC poll represented data from people answering cell phones and landlines. Who even has a landline anymore? And who answers their cell phones when they don't know who's calling and then stays on the line to answer questions from a pollster?

Do they collect age data on respondents? Because I suspect the data is strongly skewed toward older people.

[Progressive dog]

13. I only answer phones when I recognize the number

If it's important they can leave a message. I have NoMoRoBo on my home phone plus I have dozens of numbers blocked. On my cell phone sometimes I use Google to screen calls but usually I let the call go to voice mail.
FYI I'm a senior, many of us have learned how to avoid spam too and of course that means we also avoid polls.

[yardwork]

14. Me too, and I'm also a senior.

So again I ask, who is answering these polls?

[Progressive dog]

16. Lonely people?

Would they be more likely to be Republican?
The polls do come with an error percentage which is important but seldom discussed.

A margin of error of plus or minus 3 percentage points at the 95% confidence level means that if we fielded the same survey 100 times, we would expect the result to be within 3 percentage points of the true population value 95 of those times.

https://www.pewresearch.org/short-reads/2016/09/08/understanding-the-margin-of-error-in-election-polls/
Of course the margin of error is based upon the number polled and the pollster,s calculation of how accurately he can model the average behavior of real voters based upon his sample.

[yardwork]

17. Lonely people, or people who are a little paranoid and think they need to answer every call?

Fox News appeals to people who are more fearful, more likely to believe conspiracy theories.

I don't know, I just think that phone polls could have very skewed data these days.

[ProfessorGAC]

7. A Few Things

The sampling stratification is an issue.
In the ABC poll, there may have been an abundance of democratic further left who disapprove of Biden not fixing immigration, not raising taxes on the rich, not promoting a wealth tax, not decriminalizing pot, etc.
If that happens dem leading poll responders would be "disapprove" but they'd still vote for him.
If Rasmussen were to have a different stratification, they could preponderantly have more centrist voters who would approve of his presidency.
With all thar said, 15% seems an awfully big delta.

[LanguageLover]

11. That's what I meant.....

That's what I meant. A 15% delta between any two polls is an awfully large delta. Somewhere, something is not right, even taking into account the extreme left Dems who disapprove of Biden and the like.

Your points are valid though. There are a lot of Dems who live in their own dream world of perfection and thus not seeming to like anything. If they are the ones willingly tilting polls like that of ABC, then it is high time they woke up to reality.

[Sneederbunk]

9. It's not November 5, 2024.

[intheflow]

12. Who and how are they polling?

Are they mostly polling old people on land lines? Are they sending poll invitations via email to a population that hates and ignores emails? Where are they getting aggregate info on how to contact people (like, from an NRA database of likely voters)? Each of these questions could help answer the OP's question.

[uponit7771]

18. ***LARGE MOE AND SMALL SAMPLE RATES*** M$M uses these polls as clickbait now because ...

... people don't look at the MOE being 3 or greater in 2023 when 1 - 2 MOE should be the standard.

[Deminpenn]

19. Yawn

Don't pay attention to any polls that are over a year away from the 2024 election. They are meaningless.

[ananda]

20. You mean the gamed rightwing ones?

???