National Polls - Inside the margin...

[UsualName]

Your logic is spot on.

But... 4 years ago Nate Silver showed you could forecast a chance of victory for Trump if he was within 2 to 3 percent of Hillary in the national vote. Because she had excess voters in CA and NY.


I presume the same situation still applies. Biden has a "wasted" 1 to 3 percent of the vote due to excess voters in blue states.

So the way the jem algo sees it.... Biden needs to be up by 5 to 7 percent to be outside the margin in an honest poll to be in the lead outside the margin for error.
When you look at polls with an honest turnout model... (1 to 5 more democrats and a proper sampling of independents) ... the race is inside the margin for error... or too close to call.

Hence at the moment it is as you said... its coming down to few swing states...


Even more specifically the outcome probably comes down to "cross over" voters and swing voters in the swing states.

First off kudos to you for making sense in most of this.

Second, sampling should be randomized then weighted, not a predetermined weight prior to sampling. That’s basic statistical mathematics. Saying I want 45 democrats 45 republicans and 10 independents for this poll would ensure inaccuracy.

Broadly speaking, this race will tighten and tighten into Election Day.

 

[smallfil]

I'm gonna pull a Here4money and call this a false flag op by the Democratic Party.

Democrats are getting desperate that is why. If Joe Biden is widening his lead and clearly, clobbering President Donald Trump, why would Democrats engage in all sorts of gimmicks to try to win the election? Why would they bother to try cheating on a mass scale as Ilhan Omar has shown in Minnesota? If you were winning on a mass scale as Democrat extreme liberal hacks want everyone to believe, the last thing you would want is for the election to be disrupted by mass election fraud, necessitating the US Supreme Court to intervene? You would want the elections fair and honest as you are going to win outright. However, that is the lie. You try to cheat on a massive scale because you know in reality, that you are losing and losing by a huge margin.

 

[jem]

Thanks for that...

But, we not been talking about how to start a proper poll from scratch.
I have been talking about how to unskew a crooked poll if they have slanted their poll to be democrat or republican rich.

Sure... they could also screw with independents but we are not going that deep.

First off kudos to you for making sense in most of this.

Second, sampling should be randomized then weighted, not a predetermined weight prior to sampling. That’s basic statistical mathematics. Saying I want 45 democrats 45 republicans and 10 independents for this poll would ensure inaccuracy.

Broadly speaking, this race will tighten and tighten into Election Day.

 

[TreeFrogTrader]

First off kudos to you for making sense in most of this.

Second, sampling should be randomized then weighted, not a predetermined weight prior to sampling. That’s basic statistical mathematics. Saying I want 45 democrats 45 republicans and 10 independents for this poll would ensure inaccuracy.

Broadly speaking, this race will tighten and tighten into Election Day.

We know that, duh. Always have.

You and your buddy Tony Abdul El Stark are the ones who posted polls showing massive Biden leads day after day while secretly going full-Toobin as you post them.

But then as the election nears you start sidewinding and crabwalking sideways to hedge everything you say.

Reality setting in can sober a person up.

As I have said from early on: "the election will be tighter than a bull's arse at fly swatting time." If you want to run a couple IBM mainframes for several hours to make a more scientific projection, go for it. Comes out the same.

 

[UsualName]

States in order of importance:

Trump:
Florida
Pennsylvania
North Carolina

Biden:
Michigan
Wisconsin
Minnesota

Thanks for that...

But, we not been talking about how to start a proper poll from scratch.
I have been talking about how to unskew a crooked poll if they have slanted their poll to be democrat or republican rich.

Sure... they could also screw with independents but we are not going that deep.

So as to your skewing the skew, just go look at it in the poll. Every pollster has a different electorate weighting based on expectations. However, this year has a lot of reasons why some very smart pollsters can be very wrong. Namely:

1. increased mail in voting
2. With additional mail in voting comes more opportunity on Election Day get out the vote initiatives
3. Increased voter registration (Republican specifically)
4. Covid, Covid, Covid
5. Much Higher than normal enthusiasm

What I will say is Senate races and state polls are good indicators of national polling accuracy.

Senate races are showing bad Republican senate candidates may actually be driving some polling down for Trump, such as in Arizona, Colorado and Iowa, and relatively stronger dem candidates in these states may be driving polling up for Biden.

The real question here is can the middle of the ticket carry the top of the ticket, in say Arizona or Iowa.

State polls are showing a much closer race, like you originally stated - Trump is actually a normal polling error away from tied or leading in Pennsylvania and Florida. If Pennsylvania is off that would mean a lot of the Midwest is off too.

 

[jem]

First of all nice balanced post.

I noticed that a recent poll by a national polling Company found that only 1% more people identified as democrat over republican....
that is a big change over past presidential elections... if I recall.
You covered that with your post regarding R registration... sort of. Nicely done.

2. I agree with your other points regarding the variables... and that is why... I give a large range to what the Algo considers an honest poll... 1 - 5 more democrats and a reasonable sample of independents.

States in order of importance:

Trump:
Florida
Pennsylvania
North Carolina

Biden:
Michigan
Wisconsin
Minnesota


So as to your skewing the skew, just go look at it in the poll. Every pollster has a different electorate weighting based on expectations. However, this year has a lot of reasons why some very smart pollsters can be very wrong. Namely:

1. increased mail in voting
2. With additional mail in voting comes more opportunity on Election Day get out the vote initiatives
3. Increased voter registration (Republican specifically)
4. Covid, Covid, Covid
5. Much Higher than normal enthusiasm

What I will say is Senate races and state polls are good indicators of national polling accuracy.

Senate races are showing bad Republican senate candidates may actually be driving some polling down for Trump, such as in Arizona, Colorado and Iowa, and relatively stronger dem candidates in these states may be driving polling up for Biden.

The real question here is can the middle of the ticket carry the top of the ticket, in say Arizona or Iowa.

State polls are showing a much closer race, like you originally stated - Trump is actually a normal polling error away from tied or leading in Pennsylvania and Florida. If Pennsylvania is off that would mean a lot of the Midwest is off too.

 

[UsualName]

First of all nice balanced post.

I noticed that a recent poll by a national polling Company found that only 1% more people identified as democrat over republican....
that is a big change over past presidential elections... if I recall.
You covered that with your post regarding R registration... sort of. Nicely done.

2. I agree with your other points regarding the variables... and that is why... I give a large range to what the Algo considers an honest poll... 1 - 5 more democrats and a reasonable sample of independents.

Thanks but I don’t try to be balanced. I’m actually opposed to people who try to balance issues for the sake of balance. That in itself creates an inaccuracy. If Republicans have a registration gain, it is what it is.

 

[Nine_Ender]

Thanks but I don’t try to be balanced. I’m actually opposed to people who try to balance issues for the sake of balance. That in itself creates an inaccuracy. If Republicans have a registration gain, it is what it is.

I think it's been obvious all along Trump is losing and if anything it's gotten worse as he acts mentally ill in the campaign. People are trying to dream up a contest that isn't going to happen; even the Republicans know they've lost a ton of female votes with Trump's horrible management of the virus ( among other reasons ). Just on the female vote alone, it's a done deal. If the minority vote gets enthused about this election, and the evidence suggests they are, even more reason Democrats will win.

Guess we'll see but I doubt the final result justifies these ideas that it's a close call. Let's see if Trump gained ( unlikely ) or lost votes in these two categories. Unlike this forum, half the population is female.

 

[noddyboy]

I think it's been obvious all along Trump is losing and if anything it's gotten worse as he acts mentally ill in the campaign. People are trying to dream up a contest that isn't going to happen; even the Republicans know they've lost a ton of female votes with Trump's horrible management of the virus ( among other reasons ). Just on the female vote alone, it's a done deal. If the minority vote gets enthused about this election, and the evidence suggests they are, even more reason Democrats will win.

Guess we'll see but I doubt the final result justifies these ideas that it's a close call. Let's see if Trump gained ( unlikely ) or lost votes in these two categories. Unlike this forum, half the population is female.

What do you make of this then?

https://www.zerohedge.com/markets/j...the+survival+rate+for+everyone+drops+to+zero)

 

[jem]

Facts and data have no useful correlation with nine moron's firmly stated opinions. (technically, there is frequently an inverse correlation.)

What do you make of this then?

https://www.zerohedge.com/markets/jpmorgans-kolanovic-has-another-warning-those-expecting-crushing-biden-victory?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed:+zerohedge/feed+(zero+hedge+-+on+a+long+enough+timeline,+the+survival+rate+for+everyone+drops+to+zero)