Polling isn't off. It's wrong.

[angrychair]

Polling isn't just wrong. It's broken. Two polls, by two different organizations, taken in the same time frame, one says Biden +21 with 18-34 yr olds and the other says Trump +12 with 18-34 yr olds.

I would take any polling, in our favor or not, with a hefty grain of salt.
https://www.threads.net/@mollyjongfast/post/C43SiLwuXOX/?xmt=AQGzor8oErJTBMSzdVdZhcAfFJHCbWFbMUfQBFAX8E0cdA

[wryter2000]

2. Polling tells us nothing. So do approval ratings.

Approval ratings are just polls. We have to ignore both.

The only hard data we have are vote results. Those are actual voter behavior.

Elections favor Democrats. Nikki Haley votes are votes against Trump. Those things are real.

[Chainfire]

3. Polls are bought and paid for. A well designed poll can give you what ever results you desire.

Our society has gotten to the point that you can not take anything at face value; you can't believe anyting you see, hear or read. Oh, what ever happened to the days of Walter Cronkite?

[angrychair]

6. News media was suffering financially

They decided, collectively or independently, that it was more profitable to shape the news you wanted then to leave it to chance.
They decided that misery, uncertainty and suffering sell more news headlines than peace and prosperity.
They are feeding the beast because they found out it's the beast that lays the gold eggs, not the goose.

[WarGamer]

4. Then what about this?

Polls out of MI and WI show strong polling for Democratic Senate candidates... but Trump leading Biden.

Are you saying the Senate part of the poll is correct but the General, not so much?

Or are you saying that at this stage... March 2024... polls are irrelevant and will look different over the summer?

[Fiendish Thingy]

7. Without having seen the polls in question, my guess is

There were a large number (double digit %) of respondents choosing “undecided” or third party candidates for president, but not for senator.

[WarGamer]

11. Good point...

And we need people in the voting booth to look down and see Biden and Trump and say "No way Trump can win" and vote Biden no matter what.

People may not LOVE Biden, like they did Obama... but they know Trump is a threat.

I think election day we'll see better numbers than polling.

[angrychair]

9. I'm saying all polls

If they favor us or not. No matter the office.
There is a strong argument for selection bias, especially in more narrow age ranges that have grown up in a very different world that older polling models can accurately measure sentiment in when they don't use landlines, far more jaded and lived their lives in a world where things like abortion and gay marriage are settled in their minds.

[WarGamer]

10. yeah... that makes sense.

[Fiendish Thingy]

5. Always read the methodology and cross tabs

Because modern polling is so difficult, time consuming and expensive, pollsters often cut corners by accepting smaller sample sizes (which increase the margin of error), and deliberately or unintentionally oversample certain segments of the population (especially Republicans), and then use unscientific, non-standard “reweighting” procedures to correct for that.

In the examples regarding young voters, my guess is 1) there was an imbalance in partisan affiliation in the samples, and 2) the subgroup sample size was so small that the MOE was huge.

[ProfessorGAC]

8. Good Explanation!

I wholeheartedly agree with every part of it.

[ShazzieB]

16. Me, too.

[unblock]

12. My stats understand is pretty basic but doesn't further increase the margin of error?

I'd think that if you say hey, this portion of the pool is 6% when it really should be 12%, so you double it, aren't you amplifying any error from that portion of the pool?

[Phoenix61]

20. I have tried to explain "reweighting" to people

so many times and, for the most part, have given up. If someone doesn’t understand basic statistics they aren’t going to get it.

[Fiendish Thingy]

25. Reweighting isn't part of basic statistics

It’s an unstandardized, unscientific tool used to make a shitty sample less shitty, or in some cases, to provide the pollster with the results they were seeking.

Most reweighting methodology is unique and proprietary to the specific polling house, and rarely disclosed with any transparency in the methodology for the poll.

If you can find any detailed reweighting methodology for any major polling house, please post it here.

[Phoenix61]

27. My point was if you don't understand a random sample

which is standard stats you’ll never understand the machinations of trying to turn a not random sample into one.

[Fiendish Thingy]

30. Indeed

I would hope DU’ers, with just a little time invested, would pick up the basics of sample size and MOE.

It would save a lot of unnecessary hand wringing.

[Think. Again.]

13. I'm sure there's a premium on pollsters selling the media....

....polls that say what the media buying them want them to say.

[All Mixed Up]

14. None of that proves polling is wrong.

It just proves interpretation of the polling data is wrong. Truthfully, these polls shouldn't publish their crosstabs because crosstabs have a much larger MOE than the actual poll itself, as well a lower confidence rate.

There's no reason to dive into crosstabs because of that and yet the media does.

The only demographic polls I'd trust are polls that specifically survey just that demographic. Therefore, they have a much more reasonable MOE and confidence rate due to larger sample size.

But these polls?

If they're polling 1,000 voters and 10% are 18-34 or whatever, you're then looking at just 100 voters. That's where the noise makes it impossible to actually get a decent reading.

It's just the media failing to understand how to interpret their own data.

[4lbs]

15. Yes, let's poll a small amount of people, using technology and methods

only 2% of the country still uses primarily,

Meanwhile, the real voters use other methods.

Remember this guy called Abraham Lincoln? He was up for re-election in 1864. Polls showed he was in danger of losing badly. They said people were upset the country was at war, and a Civil War at that one. His forces weren't really winning anything, except for this one general out west (Grant) who liked to drink a lot and throw thousands of men into battle after battle, like they were in a meat grinder.

Polls said he was going to lose re-election. Badly.

What happened? He easily won the popular vote and demolished his opponent electorally.

Yup. Sure lost badly didn't he?

Also, remember 1964 is the dividing line politically. 1864 Republicans are like current Democrats.

So, while the GOPpies like to bleat about "Lincoln created the GOP." Umm... yeah. But back then, the GOP was like the current Democratic party. Watch them try to hem and haw their way out of that one.

It wasn't until 1964 that all the white racist Democrats joined the GOP and formed the current white racist Republican party.

So, before 1964.... GOP = 2000s Democrats, and Democrats = current GOP.

After 1964... current Democratic party, and current batshit crazy white racist GOP.

[NoMoreRepugs]

17. They are mostly garbage.

[GiqueCee]

18. Pollsters...

... manipulate responses by the way they phrase their questions, so as to get a response that whoever commissioned the poll wants to hear.
And the responses are also slanted by where they choose to ask those questions. If they park their asses in diner smack dab in the middle of a crimson county somewhere in the Midwest, the results will be predictable, but certainly not indicative of the country as a whole. It's all bullshit meant to steer voters toward the outcome preferred by frothing-at-the-mouth sociopaths like Leonard Leo, or the sole surviving Koch brother. They're all rotten down to their DNA.

[Phoenix61]

21. My favorite example of a loaded question is...

The poll stated it showed most people didn’t believe the Holocaust happened but didn’t include the question. I found the question in another publication.
“Do you believe it’s not possible the Holocaust didn’t happen” yes or no answer required.
I’m still not sure how to answer it to show I do believe it happened.

[GiqueCee]

22. BINGO!

Exactly the sort of semantic skullduggery I was referring to.

[EarthFirst]

24. Many younger participants aren't even reading many of the questions...

“For example, in a February 2022 survey experiment, we asked opt-in respondents if they were licensed to operate a class SSGN (nuclear) submarine. In the opt-in survey, 12% of adults under 30 claimed this qualification, significantly higher than the share among older respondents.”

[Warpy]

19. It's also irrelevant at this point in an election year

since it's all about name recognition among people whose recent memory extends only as far back as the last time they saw any news on TV. It's about the whim of the moment, not about a weighty decision they will make when they do vote. It's about how they feel, not about what they might think, if they do bother to think beyond their own lives and problems.

That's why polling at this point is always going to be wrong.

[Mr.Bill]

23. We DUers live in what I would call

a bubble of information. I mean reliable information. We loose sight of the large % of the population who are misinformed or uninformed. They exist in all demographic groups, political parties, income bracket and education level.

I swear my parents (blue collar and not well educated) would vote for someone because "they have a good American-sounding name". These people exist still, and they respond to polls.

[miyazaki]

26. They can all poll my finger. nt

[elocs]

28. If the polls overall are NOT off, wrong, or broken and Trump wins

then there will be a lot of people with egg on their faces and wondering what the hell happened.
But if the polls are off, wrong, or broken then how will they ever recover any credibility again?

[DontBelieveEastisEas]

29. I think the polling is wrong.

Last edited Sat Mar 23, 2024, 09:27 PM - Edit history (4)

I have edited everything I wrote the first draft, because I thought the numbers had an obvious flaw, now I tend to agree the poll is just wrong,

A January poll by this outfit had some similar fishy numbers.
Total, 18-34, 35-49, 50-64, 65+

Joe Biden, Democrat
44%,,,,,,, 43%, 37%, 45%, 52%
Donald Trump, Republican
44%,,,,,,, 41%, 50%, 47%, 40%

Which also shows the oldest age group being strong Biden supporters.

You will notice in the OP that the same poll that has Biden behind with young voters, shows him ahead with older voters.
I thought, the numbers were likely accurate, but flipped. (Biden's counts are Trump's counts)

But, it seems the poll really does report this.

It seems the thread's post is accurate from the link at,
https://civiqs.com/documents/Civiqs_DailyKos_monthly_banner_book_2024_03_8zn7pf.pdf

I had believed it was transcribed in error!

Research Methodology
https://civiqs.com/methodology

It's given, I believe, 2.5 stars out of 3 on 538, but I don't trust it from here on out.

[Silent3]

31. I hope the polling is wrong, but we're beginning to sound like a bunch of people whistling past the graveyard here

DUers clearly see how fucked up Trump is, and how dysfunctional the Republican party has become. But I think too many of us have too high an opinion of American voters.

All this talk of it "being too early" for voters to be "tuned in" and for polls to matter?

How tuned in do you have to be to miss the ranging dumpster fire that is Trump and the Republican party?

For those who can't and don't see it now, I'm not sure how much fucking waking up such people are going to do over the next few months.

For me, the big hope is that Trump is going to crumble under the pressure of recent events and his declining mental faculties, and this decline will (if it happens severely enough, quickly enough) finally be enough to wake up the ignorant sleepwalkers in our midst.