The Aug-Sept Poll on Latino's in Florida has 4.5% MOE!! WTF Media? You guys did this shit in 16 !!

[uponit7771]

I just saw the MOE on Steph Ruhles show, it is 4.5% !! Its an LV poll but still the MOE is high even though the amount of people were just a slice higher than 750 people.

This is the Marrist Polling here - https://www.nbcnews.com/politics/meet-the-press/poll-trump-biden-are-tied-battleground-florida-n1239510

That poll is not as accurate relative to the other polls we have of swing states with likely voters and MOE (Margin Of Error) in the low to mid 3s (premo polling data) swing towards Biden.

That's the big difference in polling now vs in 16; likely voter polls and low MOEs in the mid 3s swing towards Biden were in 16 it was Adult and registered voters and high MOE crap that none of the media should be going off of because there could be more accurate polling.

Now the polling can get down to 2 - 3% so 4.5 to 6% isn't enough accuracy to develop a narrative off of.

Just damn, look were "Horse race" shit got us ... I'm turning every show who wants to make an narrative off these relatively high MOE polls even if they're LV polls.

Note: There are other polling that went along with the Marrist poll but there MOE was even higher. The media should DEMAND higher moe polling so they can't get viewers with "horse race" shit.

[beachbumbob]

1. why its best to ignore individual polls in the first palce

the media has their agenda and it not to make biden or democrats look good. The real crap is still coming with all the slime thrown at Biden/Harris as possible and the MSM will be all part of it

[Proud liberal 80]

2. I was watching her this morning

And turned the channel as soon as she mentioned this as the next topic, tired of hearing about this narrative.

[Drunken Irishman]

3. That MOE isn't too high at all.

It falls within the acceptable range.

Take the latest North Carolina Suffolk poll that has Biden up 3 & 4 (with and without third parties respectively) in North Carolina. The MOE for that poll is 4.4. Suffolk is a very well respected pollster.

With all that said, one poll is one poll. There's plenty other polls showing a different outcome.

[uponit7771]

4. Relative to 2 - 3% polls it is high, we should NOT accept 4+ moe polling any longer. We can get

... the polling more accurate now than even 5 years ago.

Either that or there needs to be a polling science standard, looks like we're all over the place on polling

[Drunken Irishman]

8. You're not going to find many polls with that MOE

...and if you do, they're suspicious because they've likely been polled over a period of weeks or have a ridiculously large sample size.

Most polls are going to be between 3-5 MOE.

[Fiendish Thingy]

9. Not true - many/most national polls fall between 2-3%

It’s the state polls that Often have smaller sample sizes, and thus, higher MOE’s.

Just remember that MOE’s are for the number reflecting support of a single candidate; to calculate the variance of the margin, you have to double the MOE to account for possible movement up or down.

So a poll with a 4.5% MOE is nearly useless, unless it shows a margin of 9% or greater.

[Drunken Irishman]

11. This isn't true.

YouGov's national poll: MOE 3.4/3.7
Reuters national LV model: 3.8
Monmouth national poll: 3.6

I picked just three random national polls that were released recently. All three of those had a MOE of at least 3.5 in their LV model.

Yes, states polls are not as reliable as national polls but a MOE of 4.5 is not extreme and well in line with the polls I listed.

[Fiendish Thingy]

13. Other polls MOE from this week:

Fox: 2.5%
Morning Consult: 1%
Harris: 1.6%
Rasmussen: 2%

4.5% MOE for any national poll is extreme.

[Drunken Irishman]

15. This proves my point. Thank you.

The four most disputed polls, who often have the weirdest results.

So, a lower MOE means dick. You just proved my point. Unless you think Rasmussen, with its two-point MOE, is a better poll than Monmouth. And if you think that, well, Biden is going to lose the election because he trails in that poll.

So, whose poll is right? Do you believe Trump is leading nationally?

[Fiendish Thingy]

16. Proves my point - that a national poll with a 4.5 MOE is extreme

Of all the examples we both listed, only one had a MOE over 4%.

All other things being equal, An 8 pt lead with A 4.5% MOE is a statistical tie, where a 7 pt. lead with a 3% MOE is a firm lead regardless of variance.

Yes, polling houses vary in quality, but MOE matters tremendously, and is too often ignored or overshadowed by the headline grabbing horse race of growing/shrinking leads

[Drunken Irishman]

17. Just admit you're wrong.

You stated:

most national polls fall between 2-3%

I picked three random polls that were above 3% MOE.

Your reply is to cite polls that are largely graded low by people like Nate Silver.

4.5 is not bad - especially for state polls. The fact you're hung up on this shows you don't understand how state polling works. You'll rarely find state polls which have a smaller MOE than national polls - or even similar MOE. So, calling out this poll for this MOE is ridiculous and honestly shows a lack of understanding, unfortunately.

Just take a look at the most recent state polls released today and their MOE:

Arizona (Monmouth): 4.8
North Carolina (USA Today/Suffolk): 4.4
California (PPIC): 3.9

Tell me why you'd call this poll out and not every other state poll? It's just disingenuous. And then to mention Rasmussen, which has Biden losing, and has long been a biased poll, suggesting because it's lower MOE than other far more reputable polls, is just wild to me.

Fine. You clearly think Rasmussen's poll is better than the polls that have Biden up because of the MOE.

If that Rasmussen poll is correct, it means he's likely losing among those same Latinos.

[Fiendish Thingy]

18. My point remains unchanged:

All other things being equal, An 8 pt lead with A 4.5% MOE is a statistical tie, where a 7 pt. lead with a 3% MOE is a firm lead regardless of variance.

The Monmouth AZ poll you mention has versions with Biden anywhere from +4 to tied; with an MOE of 4.8% that renders any lead in the poll nearly meaningless- if you were running Biden’s campaign, would you be excited that he’s winning in AZ? 538 has the race leaning to Biden, likely due to other polls with a strong Biden lead, including an OH Predictive insights poll giving Biden a 10 pt. Lead with an MOE of 4%.

The Suffolk NC poll gives Biden a 3-4% lead, but with a MOE OF 4.4%, that lead is not significant, it’s statistically tied, and the poll is essentially useless, and one reason why 538 has the race as a toss up.

[Drunken Irishman]

20. Your point is to dismiss a poll because the MOE is typical for a state poll...

...when it's in line with a lot of other MOE for polls. The MOE doesn't invalidate the poll because it's not a bad MOE like you suggested in your initial post:

Its an LV poll but still the MOE is high even though the amount of people were just a slice higher than 750 people.

It's not a high MOE. That's the whole argument.

[Fiendish Thingy]

21. That's not my post you quoted, it's the OP

A poll may be a valid poll from a reputable polling house, but it’s significance or utility may be limited because the high MOE essentially erases any narrow lead Biden may have.

I think you’d agree the trend is much more important than a single poll.

I do appreciate this extended back and forth on MOE’s , as it is a much overlooked topic.

[Drunken Irishman]

22. Either way, the MOE is not extreme for a state poll.

It falls well within the norm of state polling MOE and shouldn't be dismissed at any point because of it. It's not a high MOE as the OP insinuated, rather a typical MOE for a state poll.

[still_one]

5. The problem is, as I understand it, is that the republicans have been placing ads and pushing the

lie that either Biden/Harris are socialists or "controlled" by the Democratic left. That does not register well with many Cuban Floridians, however, both Biden and Harris have are visiting the state, and throwing ads not only to counter that, and also show why trump is a real threat to social security and medicare. Bloomberg is also committing 100 MILLIOIN in Florida to fight that. Bloomberg would not commit that much money to one state if he didn't think it was winnable

I think the problem with trump in Florida is that he has topped out there, while Biden/Harris have a lot of potential to make in roads there.

There are about one million Puerto Ricans in Florida, and they sure won't be going for trump

[uponit7771]

6. +1

[Cicada]

7. The typical error for a state poll has been 6 points, historically

Typical error on national polls has been 3 points typically

Don’t expect state polls to be very accurate

Though Siena state polls coordinating with the new Nate, Nate Cohn who replaced Nate Silver at NYT, was only off on average 3 points in 2018. Siena seems to have invented a better mouse trap for state polls, starting with actual registration lists in selecting people to call. A private firm somehow has phone numbers for actual voter registration people and Siena used that list.

[Wounded Bear]

10. MoE is based almost entirely on number of people polled...

To get a lower MoE, they have to poll more people. It's all basic math.

[Takket]

12. No offense but this just reads like a paranoid conspiracy theory

That the media is screwing around with 1.5% worth of error in a poll of Latinos to help make the race look closer?

I’ve seen the media do some stupid and some downright criminal bullshit but this isn’t one of them.

[uponit7771]

14. Who said anything about the media screwing with polling?!?! They're not demanding poling companies

... higher MOE that's available.

[Klaralven]

19. Sample bias and other systematic errors are probably a worse problem than MOE from sample size