NBC/Marist NC: Biden +6

[WD]

Marist is usually a good pollster, no?

Ask Sen. Donnelly and Sen. McCaskill.

Marist's final polls had McCaskill at 47% (she got 46%) and Donnelly at 45% (he got 45%).

Quote deleted by moderator

They were still wildly off margin wise. Also, if a Candidate gets a lower share of the vote than your polls show (like with McCaskill) then theirs something wrong with your polls.

[Gass3268]

The fact that 538 had Marist as a A+ pollster after there misses in 2018 is LOL

Their ratings are more based on methodology, with only a % going to results. Marist ticks off most of the what you'd want for top level methodology.

[Alben Barkley]

Marist is usually a good pollster, no?

Ask Sen. Donnelly and Sen. McCaskill.

Marist's final polls had McCaskill at 47% (she got 46%) and Donnelly at 45% (he got 45%).

Quote deleted by moderator.

They were still wildly off margin wise. Also, if a Candidate gets a lower share of the vote than your polls show (like with McCaskill) then theirs something wrong with your polls.

A 1% lower share is utterly insignificant statistically.

Being off margin wise doesn’t matter so much if you have a candidate over 50%. Trump could get all the undecideds and still lose.

[Holy Unifying Centrist]

https://twitter.com/mmurraypolitics/status/1322223323565215744?s=20

The college education rate in North Carolina is 38.1, so 40% is not off by at all.

Where are you getting 38.1% from? Wikipedia says 30% (29.9%)

https://en.wikipedia.org/wiki/List_of_U.S._states_and_territories_by_educational_attainment

[Gass3268]

https://twitter.com/mmurraypolitics/status/1322223323565215744?s=20

The college education rate in North Carolina is 38.1, so 40% is not off by at all.

Where are you getting 38.1% from? Wikipedia says 30% (29.9%)

https://en.wikipedia.org/wiki/List_of_U.S._states_and_territories_by_educational_attainment

Statistical Atlas

[new_patomic]

The 2019 census estimate is 30.5 percent of NC has a Bachelors or higher.

While the electorate probably does skew more educated than the state as a whole, 40 percent is too high.

[LimoLiberal]

Interesting disparity between this poll and internal Democratic anxiety over the state >>

https://twitter.com/jmartNYT/status/1322222512361639939

[new_patomic]

https://twitter.com/mmurraypolitics/status/1322223323565215744?s=20

The college education rate in North Carolina is 38.1, so 40% is not off by at all.

Where are you getting 38.1% from? Wikipedia says 30% (29.9%)

https://en.wikipedia.org/wiki/List_of_U.S._states_and_territories_by_educational_attainment

Statistical Atlas

Okay I see where that's coming from.

They're including people with an associates degree.

[Gracile]

https://twitter.com/mmurraypolitics/status/1322223323565215744?s=20

The college education rate in North Carolina is 38.1, so 40% is not off by at all.

Where are you getting 38.1% from? Wikipedia says 30% (29.9%)

https://en.wikipedia.org/wiki/List_of_U.S._states_and_territories_by_educational_attainment

Statistical Atlas

Should be noted that Statistical Atlas' education numbers are only for people 25 and older (which probably would be closer to what the electorate would be than simple % educated as a share of the state's population).

[Gass3268]

Interesting disparity between this poll and internal Democratic anxiety over the state >>

https://twitter.com/jmartNYT/status/1322222512361639939

You skipped this tweet.

https://twitter.com/jmartNYT/status/1322227763156291584?s=20

[redjohn]

It says a lot that I also thought this was a national poll and my heart dropped... at a *6 point lead*.

Anyways, NC is lean Biden, maybe likely.

[Asta]

The fact that 538 had Marist as a A+ pollster after there misses in 2018 is LOL

https://fivethirtyeight.com/features/weve-updated-our-pollster-ratings-ahead-of-the-2020-general-election/

I hate to agree but you're right. Marist did horribly in 2020 primaries as well.

[Pollster]

Marist is usually a good pollster, no?

Ask Sen. Donnelly and Sen. McCaskill.

Marist's final polls had McCaskill at 47% (she got 46%) and Donnelly at 45% (he got 45%).

No.  they had Mccaskill at 50 and Donnelly at 48.

They most certainly did not:

Indiana they had Donnelly 45, Braun 42, Brenton 7, Other 1, Undecided 5
Missouri they had McCaskill 47, Hawley 44, Campbell 3, Crain 2, Other 1, Undecided 3

Republicans in both appeared to benefit from third parties underperforming their polling (as they almost always do) and winning over most undecideds (common in safe states for the dominant party).

[Pollster]

Marist is usually a good pollster, no?

Ask Sen. Donnelly and Sen. McCaskill.

Marist's final polls had McCaskill at 47% (she got 46%) and Donnelly at 45% (he got 45%).

Quote deleted by moderator

They were still wildly off margin wise. Also, if a Candidate gets a lower share of the vote than your polls show (like with McCaskill) then theirs something wrong with your polls.

Margins are close to meaningless in polling, and a 1% over/underperformance is statistically insignificant unless you have a margin of error of less than 0.5% thanks to rounding, which you don't achieve until you have a sample size of ~13,000.

[TiltsAreUnderrated]

Marist is usually a good pollster, no?

Ask Sen. Donnelly and Sen. McCaskill.

Marist's final polls had McCaskill at 47% (she got 46%) and Donnelly at 45% (he got 45%).

No.  they had Mccaskill at 50 and Donnelly at 48.

They most certainly did not:

Indiana they had Donnelly 45, Braun 42, Brenton 7, Other 1, Undecided 5
Missouri they had McCaskill 47, Hawley 44, Campbell 3, Crain 2, Other 1, Undecided 3

Republicans in both appeared to benefit from third parties underperforming their polling (as they almost always do) and winning over most undecideds (common in safe states for the dominant party).

You are both right - they had two variants for the final poll, ones with named third parties and ones without. Among likely voters, the ones with named third parties had the results you've posted and the others had McCaskill 50, Hawley 47, "Other" <1 and Donnelly 48, Braun 46, "Other" 2.

[Bootes Void]

Marist is usually a good pollster, no?

Ask Sen. Donnelly and Sen. McCaskill.

Marist's final polls had McCaskill at 47% (she got 46%) and Donnelly at 45% (he got 45%).

No.  they had Mccaskill at 50 and Donnelly at 48.

They also had Evers in Wisconsin at 53 but Marist isn't a bad pollster, they weigh by region. You need to mix it up, some can weigh by region, others by education to get a good average overall

[Roronoa D. Law]

If all the blue avatars have left is (blank) was wrong in 2016 or FL 2018 then they're in trouble.

[lfromnj]

If all the blue avatars have left is (blank) was wrong in 2016 or FL 2018 then they're in trouble.

IN 2018, MO 2018, TN 2018, WI 2018 gov and to a degree FL 2018 for Marist.

[Roronoa D. Law]

If all the blue avatars have left is (blank) was wrong in 2016 or FL 2018 then they're in trouble.

IN 2018, MO 2018, TN 2018, WI 2018 gov and to a degree FL 2018 for Marist.

Oh no Biden is in trouble of losing MO, TN, and IN.

[Hope For A New Era]

Marist is usually a good pollster, no?

Ask Sen. Donnelly and Sen. McCaskill.

Marist's final polls had McCaskill at 47% (she got 46%) and Donnelly at 45% (he got 45%).

^^ important to note

[President Johnson]

Insane margin. I guess it will be closer, especially since this poll has almost a 5% margin of error.

[Gone to Carolina]

As long as they refuse to weight by education I will not trust Marist.

[philly09]

As long as they refuse to weight by education I will not trust Marist.

Pretty much this.

[ProgressiveModerate]

As long as they refuse to weight by education I will not trust Marist.

Very good point. This poll seems a bit too bullish on Biden, but it does cancel out Trafalgar at least.

[bandg]

Marist in 2018 was truly horrendous across the board.

Donnelly +2, actual Braun +6 (8 point error)
McCaskill +3, actual Hawley +6 (9 point error)
Evers +10, actual +1 (9 point error)
Dewine Tied, actual +4 (4 point error)
Gillum +4, actual Desantis +0.4 (4 point error)
Nelson +4, actual Scott +0.2 (4 point error)
Sinema +6, actual +2 (4 point error)