Biden approval ratings thread, 1.0

[Annatar]

A lot of these Biden approval polls are oversampling democrats, the CNN poll that had him at 51% approve had a D+6 sample, no election since 2012 has had an electorate that democratic, the Marist poll that has Biden at 49% approval has a D+7 sample, 34% democratic, 27% Republican, for multiple cycles now we just see pollsters continually oversampling democrats relative to what the exit polls show, is Biden really at 49% approval, with a D+7 sample he might be, in reality he is likely lower.

[Annatar]

A lot of these Biden approval polls are oversampling democrats, the CNN poll that had him at 51% approve had a D+6 sample, no election since 2012 has had an electorate that democratic, the Marist poll that has Biden at 49% approval has a D+7 sample, 34% democratic, 27% Republican, for multiple cycles now we just see pollsters continually oversampling democrats relative to what the exit polls show, is Biden really at 49% approval, with a D+7 sample he might be, in reality he is likely lower.

the NPR poll has Biden's disapproval at nearly 20% among Blacks and only 51/43 among Latinos. That seems... very unlikely

Sub group cross tabs in polls are always weird, not sure what can be done about it.

[Annatar]

A lot of these Biden approval polls are oversampling democrats, the CNN poll that had him at 51% approve had a D+6 sample, no election since 2012 has had an electorate that democratic, the Marist poll that has Biden at 49% approval has a D+7 sample, 34% democratic, 27% Republican, for multiple cycles now we just see pollsters continually oversampling democrats relative to what the exit polls show, is Biden really at 49% approval, with a D+7 sample he might be, in reality he is likely lower.

the NPR poll has Biden's disapproval at nearly 20% among Blacks and only 51/43 among Latinos. That seems... very unlikely

Sub group cross tabs in polls are always weird, not sure what can be done about it.

So it's best to take a poll at face value and not try to unscramble it's crosstabs.

Edit: Admittedly, I'm very guilty of doing this.

Party ID and education are 2 things that do need to reflect the electorate though just like the number of men vs women in a poll or the entire poll gets thrown off.

[Annatar]

A lot of these Biden approval polls are oversampling democrats, the CNN poll that had him at 51% approve had a D+6 sample, no election since 2012 has had an electorate that democratic, the Marist poll that has Biden at 49% approval has a D+7 sample, 34% democratic, 27% Republican, for multiple cycles now we just see pollsters continually oversampling democrats relative to what the exit polls show, is Biden really at 49% approval, with a D+7 sample he might be, in reality he is likely lower.

the NPR poll has Biden's disapproval at nearly 20% among Blacks and only 51/43 among Latinos. That seems... very unlikely

Sub group cross tabs in polls are always weird, not sure what can be done about it.

So it's best to take a poll at face value and not try to unscramble it's crosstabs.

Edit: Admittedly, I'm very guilty of doing this.

Party ID and education are 2 things that do need to reflect the electorate though just like the number of men vs women in a poll or the entire poll gets thrown off.

While this is true for education, party ID is fluid; attempting to weight on it is fraught with problems.  Party ID is not a demographic characteristic.

Party ID at the national level is pretty stable, 2016 was D+3, 2018 was D+4, 2020 was D+1, polls that vary to much from that need to be looked at with scrutiny.

[Annatar]

I remember many democrats over the past decade claiming non college whites in Iowa will never vote like non college whites in the south because they are less religious, less racist, more liberal etc, if Biden really is in the low 30’s in Iowa, non college whites in Iowa might start voting like southern whites.

[Annatar]

I remember many democrats over the past decade claiming non college whites in Iowa will never vote like non college whites in the south because they are less religious, less racist, more liberal etc, if Biden really is in the low 30’s in Iowa, non college whites in Iowa might start voting like southern whites.

Biden did win 39% of the non college white vote in Iowa, this is much much higher than anywhere in the south

Yes he did, that is why him being at 31% approval implies that 2024 might be when non college whites in Iowa vote like southern ones.

[Annatar]

https://twitter.com/PollWatch2020/status/1458572358487355395?s=20

This…. Had trump up in Pennsylvania

Their poll had Trump up 1% in PA, he lost by 1.2%, they missed by just 2%, they were more accurate than most pollsters in PA that had Biden up 5% and missed by 4%.