Polling: Unskewed Polls and comments on polling (moved from Harris Thread)

Is it your understanding that the NYT is only read in the NY subs???

I mean your point of who it does tend to reach not being who she needs to reach is actually valid, but that statement is either a whoosh or something else.

Actually I could see a live uncut Fox interview doing her good. Yes it would a hostile interviewer. Her holding her own in that circumstance reaches more of who she needs to reach than a NYT interview would.

Agreeing with this post ^.

Where did i say that?

According to The New York Times readership demographics, 91% of its readers identify as Democrats.

I would say 91% is “solidly blue”. Do you disagree?

Not long ago Gavin Newsom held his own in an interview with Hannity. I think Harris could do the same.

Yes, Harris could hold her own on a Fox interview. Until they cut and edited and made some slight little changes to make her look bad.

I misread your “subs” as suburbs; you apparently meant subscribers. Which is not crazy.

Which is why @DSeid said, and I agreed with, ‘live uncut interview’.

Do you trust Fox news? I do not. And even if they aired it once liek that, they could air it over and over again, cut to shit and gone.

But once they did that, it’s out there for every other media outlet to analyze and rebroadcast. If Fox tried some BS with cutting and editing, it would be discovered rather quickly.

I agree that would be hard for FOX to get away with.

But Harris really needs to ramp up the interviews and maybe even a press conference. She has only done the CNN thing and a radio interview with Univision’s audio division.

I don’t fully understand why she doesn’t do more interactions like this but I assume there is a reason. I know other President’s have been criticized for their lack of unscripted press encounters but I don’t get it.

And, so? So people would say Fox lied. We say that every day here. It is a known fact.

They take up days of valuable campaigning time- prep, set up, etc. trump does no interviews.

I’m not sure about the DJT claim. He seems more than willing to sit down any where, any time to blather his nonsense.

He calls in to radio shows, morning shows, sits for interviews, blah blah blah.

And the time off the campaign trail is paid back by the media/online exposure. There would be clips all over Tictok ore whatever the kids are glued to.

An interview is where the person asks the important person questions, and they answer them. Does trump answer question? No. He just blathers on. Which is why he doesnt need prep time.

Well that’s actually the problem, isn’t it?

DJT can blather on and on with his nonsensical word salad answers.

Harris can’t miss a single detail or the media (NYT mainly) will call her out for her confusion, incompetence and lack of preparation for the office.

Post debate polls are highly likely to be very negative for Harris because he will be louder, more aggressive and there will be no realistic way for her to refute all of his lies.

I think this is the post for polling discussion, so I wanted to drop an observation I made this morning and see if the gurus here had any thoughts.

Lots of poll watchers make a relatively big deal about how while Harris has a small lead in the polling averages, her lead is significantly smaller than Biden’s was at this time 4 years ago. For example, RCP has her at +1.3% while Biden was at +7.8%. This is taken to be a bearish sign for Harris.

However, when you actually look at polls from 2020 one thing jumps out - the number of “undecided” or “unallocated” voters was much higher. In early September 2020 the averages were roughly Biden 50% Trump 43%, with 7% for third party or undecided voters.

Today they have Harris 48.5% Trump 47.2% with only 4.3% unallocated. Some polls (Harvard-Harris for example) literally have no undecided voters at all - it has a 50/50 tie.

Some speculation:

  1. Voters are much more set on their choices and there really are fewer undecided voters. In theory this would reduce the likelihood of a polling error, or at least make it a pollster-weighting issue not a “undecided voters breaking strongly one way” error.
  2. Pollsters are only getting responses from decided voters. Even the “independents” answering are much more likely to have a preferred candidate than in the past. This would possibly increase the likelihood of polling error as it would indicate a bad sample. One can speculate either way about how that unpolled sample leans.
  3. 47% is Trump’s hard cap and all you are seeing in polling is how many Harris supporters (or anti-Trump voters) happen to respond. Trump got basically 47% in 2020, and 46% in 2016.

Anyway, just wanted to throw that observation out there. It seems very odd to me that this year, with a much more significant third-party candidate on the ballot in many states (even if he has officially dropped out) that there would be such a difference in the unallocated vote share from 2020.

Interesting. I think it’s mostly your 1 and 3; that is, “never-Trumpers” are slightly more numerous, and more certain of their convictions, than before. In the run-up to 2016 and 2020, some were polling as “undecided,” and in the end a fair number of them ended up not voting at all (especially in 2016, in key locales).

This time, they’re more willing to say “Harris” straight-up.

That might sound like a good thing for Harris, but it isn’t. The percentage of “yes-Trump-come-hell-or-high-water” is a bit higher than ever – 47% – and it’s not going to change. So, as you pointed out, there are very few truly undecideds left. But at least the turnout efforts for Harris shouldn’t be quite as challenging as they were for Clinton or Biden.

My complete WAG is that more than one is at play.

I do believe that 48% is pretty much his ceiling. And that his true floor is not much below that. Maybe some broke late to Trump last time, but I seem to recall post hoc analysis saying it did not seem so.

I don’t think we really know Harris’s floor or ceiling, but I suspect her floor is near Biden’s when he dropped, and that she is very near her ceiling.

I suspect that the biggest difference then and now is how the polling companies are weighting their data sets trying to correct for systemic errors in the last cycles. If so it is an unknown if the correction is on the money or off one way or another. (Or even several sorts of errors!)

Your posts nails it.

Specifically, the demographic model of who the voters in the November 2024 election will be. That is, who is actually going to be cast ballots compared to who is responding to polling queries.

Is there any kind of metric which can take the change in a national poll and approximate what the relative change will be in a specific state? For instance, say that at the last national poll, Ohio is 44% Harris, 53% Trump. If the newest national poll shows an increase of 1% to Harris from the last national poll, is there a statistical correlation to how the polls in Ohio would be expected to change?

That’s one of the things the Nate Silver’s model does. The details are proprietary.