Welcome to DU! The truly grassroots left-of-center political community where regular people, not algorithms, drive the discussions and set the standards. Join the community: Create a free account Support DU (and get rid of ads!): Become a Star Member Latest Breaking News Editorials & Other Articles General Discussion The DU Lounge All Forums Issue Forums Culture Forums Alliance Forums Region Forums Support Forums Help & Search

DLCWIdem

(1,580 posts)
Sat Mar 9, 2024, 04:28 PM Mar 2024

ANYBODY CHECK OUT THE NYT/SIENNA POLL

I know that there is criticism of the poll because it over sampled rural areas. But does anybody know what that means for weight?

8 replies = new reply since forum marked as read
Highlight: NoneDon't highlight anything 5 newestHighlight 5 most recent replies
ANYBODY CHECK OUT THE NYT/SIENNA POLL (Original Post) DLCWIdem Mar 2024 OP
It's a deeply flawed poll Fiendish Thingy Mar 2024 #1
Why would I bother? It is a shit poll that deliberately samples more Trump voters... Demsrule86 Mar 2024 #2
Making the decision to oversample R voters NanaCat Mar 2024 #3
Maybe, but the way they handle the data was completely stupid. Demsrule86 Mar 2024 #4
Thank you for linking the poll DLCWIdem Mar 2024 #8
I'm sure the tyrants love it DoBW Mar 2024 #5
These polling outfits need to go get stuffed .... LenaBaby61 Mar 2024 #6
Disproportionate sampling typically gets addressed by weighting 0rganism Mar 2024 #7

Fiendish Thingy

(21,864 posts)
1. It's a deeply flawed poll
Sat Mar 9, 2024, 04:52 PM
Mar 2024

Not only did it oversample rural areas, it oversampled southern states (35%) and oversampled Republicans and undersampled Democrats (28%, when Dems represent 38% of registered voters).

Sample size of just 980, with a MOE of +/- 4%, when most pollsters aim for sample of 1500-2000 for a MOE of 3% or less.

The simplest way to know this poll was flawed was that it claimed Trump was winning White suburban women by something like 5-7 points!

Demsrule86

(71,465 posts)
2. Why would I bother? It is a shit poll that deliberately samples more Trump voters...
Sat Mar 9, 2024, 05:04 PM
Mar 2024

it Generally uses registered voters not likely voters.
Also it assumes the following which makes this and other such bad polls

"Red Flag #1: Trump Keeps 97% Of His Supporters"

"Red Flag #2: Trump And Biden Are Tied With Women
This poll arrived in the same week when Trump announced his support for a national abortion ban. Trump brags about taking away reproductive healthcare rights from women, and his party is struggling as Alabama banned IVF, and that is just one issue. Trump was also found liable for rape. Donald Trump isn’t going to split the vote among women with Joe Biden.
Are you kidding me? After Roe and after Trump took credit for ending it, he is tied with some voters? It's idiotic on the part of these pollsters."

"Red Flag #3: Trump More Than Doubles His Support With African-American Voters.
The New York Times poll also found that there is going to be a historic African American voter realignment. The poll found that 23% of African Americans support Trump.
If 26% of black voters voted for Trump, it would be the most radical realignment in African American voting since the Civil Rights Act was passed"

https://tcinla757.substack.com/p/why-the-new-nytsiena-poll-is-bullshit

Now I don't (and the author too) think this is evidence of bias, but of bad methodology. The way they sample the data is faulty. They use landlines, and 9 times as many cell phones to correct for the use of cell phones. But the issue is only 19% of folks answer calls on cell phones where they don't recognize the name. Please consider, that the only way this poll can indicate Trump is leading is that every Trump voter will vote again for him ( we know that is not true for several reasons...one being Haley voters say they won't vote for Trump. Also Trump is not tied with Joe Biden in terms of support among women voters after Roe and Trump has not doubled the support of black voters.

This is why I call the poll bullshit as do many others. This sort of polling is why we win even though their shitty polls say otherwise. This is why we had a red fizzle and not red wave which polls expected...this is why we have won every special in 23 and 24. And this is why Biden will beat Trump handily in November.

 

NanaCat

(2,332 posts)
3. Making the decision to oversample R voters
Sat Mar 9, 2024, 05:42 PM
Mar 2024

Undersample D voters and oversample one region of the country isn't a problem of methodology. It's a problem of deliberate bias. Reputable pollsters know what the breakdowns are. Ignoring those parameters is a deliberate act.

V unprofessional.

Demsrule86

(71,465 posts)
4. Maybe, but the way they handle the data was completely stupid.
Sat Mar 9, 2024, 05:45 PM
Mar 2024

So is it on purpose or just because of their false assumptions...they over sample the rural voters because there a less of them. It is part of how they do their stupid polls. It is a bad poll by all measures.

DLCWIdem

(1,580 posts)
8. Thank you for linking the poll
Sun Mar 10, 2024, 01:18 PM
Mar 2024

I know it's funny but I still haven't got the knack even though I've been here for years

LenaBaby61

(6,991 posts)
6. These polling outfits need to go get stuffed ....
Sat Mar 9, 2024, 06:14 PM
Mar 2024

Inside the polling .....

Many polling outfits usually always over-sample older, right leaning white voters with landline phones, their MOE in those polls is high +/- 3.5-4.0, or they use a very small sampling of lets say 500-1,500 people, and those usually 'polled' are mentioned above. Few to no young cell phone users, few to no minorities polled--l think you get the picture. Its one big horse race narrative polling.

A friend whose done yougov polling for 11 years for pay, and while she won't tell me specifically about what she's polled about she said that in the political polling yougov groups Dems/GQP together as in how do you rank Congress performance in doing its job, and she told me that the option puts the blame equally on Dems/GQP. Its the Putin party who are lazy, obstructing, who want to derail commonsense legislation--not Dems. My friend told me that she tells them constantly at the end of the political survey's to please stop grouping the two parties together, and to make polling judging each political party separately on congressional/senate performance. She told me that while they 'occasionally' separate job performance between D's & GQP, they still do to much 'group blame' on why things aren't getting done in the senate/congress politically. Not saying that Dems shouldn't GOTV, phone bank, help register voters, sharpen their messaging, and a ton of other things to help Pres. Biden to SAVE our democracy, but I'm fucking sick of FAKE mostly corporate polling still asking/pushing the "ls Biden too old" shit, when shitty pants tRumputin is sweating, slurring, smelling like shit, glitching, rapist, racist lying, losing his train of thought, is still in what mind he has left running vs Pres. Obama during his embossing kkklan rallies.

And there's this 👇🏻👇🏻

538 drops Rasmussen Reports from its analysis

The poll aggregation and evaluation site 538, part of ABC News, dropped the right-wing polling firm Rasmussen Reports from inclusion in its polling averages and forecasts.

The decision comes after months of consideration that broke into public view in June. At that point, G. Elliott Morris, ABC News’s editorial director of data analytics and 538 lead, presented Rasmussen with questions meant to evaluate its objectivity and methodology. Rasmussen published the letter on its website, triggering backlash against 538 in right-wing media — and by Nate Silver, the founder of what was then called FiveThirtyEight. No change was implemented.

https://www.washingtonpost.com/politics/2024/03/08/rasmussen-538-polling/

0rganism

(25,450 posts)
7. Disproportionate sampling typically gets addressed by weighting
Sat Mar 9, 2024, 06:16 PM
Mar 2024

The trick lies in how the weights are distributed . That's when undersampling a demographic bites the poll weighting models in the butt, as a too-low number can have larger-than-expected errors that magnify when weighting multipliers are applied. This in turn leads to surprising clickbaity results for particular subgroups of the surveyed population.

Latest Discussions»General Discussion»ANYBODY CHECK OUT THE NYT...