TIPP: MI - Biden +13, FL - Biden +11

[Holy Unifying Centrist]

Michigan poll is junk
54% was college educated and only 18% have never been to college.
Florida is too
Almost 60% educated

Lmao are people actually taking these polls seriously because 538 said A/b?
Its just like Marist had a A+ lmao.

Good points. What would be the results if they were weighted properly by education?

[Holy Unifying Centrist]

Michigan-yes.

Florida-HELL NO!

I think you are being biased and not fact-based here.

The Michigan poll is actually far more inaccurate than the Florida poll here. In the Michigan sample, they have it as 60% college educated (really inaccurate; it was 37% educated according to the 2018 exit polls). This is a huge problem because educated voters in Michigan vote for Biden by 17, but non college educated are far less pro Biden. This poll is actually super skewed in favor of Biden because of this. I would say there's at least a 7% pro Biden skew because of their poor weighting on education.

I'm making a big deal about this because the lack of education weighting was the biggest contributor to why midwestern polling was so bad in 2016.

Meanwhile, the Florida poll holds up much better because there is not a significant educational gap in how people are voting. It's still way over sampling educated people, but because there isn't a big gap, it doesn't skew the results as much. It's probably only a 3-4% pro Biden skew.

Just because FL was close in 2016 and 2018 does not mean it will be close in 2020. Every election is different.

[Holy Unifying Centrist]

Michigan-yes.

Florida-HELL NO!

I think you are being biased and not fact-based here.

The Michigan poll is actually far more inaccurate than the Florida poll here. In the Michigan sample, they have it as 60% college educated (really inaccurate; it was 37% educated according to the 2018 exit polls). This is a huge problem because educated voters in Michigan vote for Biden by 17, but non college educated are far less pro Biden. This poll is actually super skewed in favor of Biden because of this. I would say there's at least a 7% pro Biden skew because of their poor weighting on education.

I'm making a big deal about this because the lack of education weighting was the biggest contributor to why midwestern polling was so bad in 2016.

Meanwhile, the Florida poll holds up much better because there is not a significant educational gap in how people are voting. It's still way over sampling educated people, but because there isn't a big gap, it doesn't skew the results as much. It's probably only a 3-4% pro Biden skew.

Just because FL was close in 2016 and 2018 does not mean it will be close in 2020. Every election is different.

Where are you getting this Michigan education numbers you are citing?

This is the education distribution of the Michigan poll:
HS or less: 18%
Some college: 29%
College Grad: 38%
Grad Degree: 16%

This is the education distribution of the 2018 exit poll:
https://www.cnn.com/election/2018/exit-polls
HS or less: 23%
Some college: 25%
College Grad: 35%
Grad Degree: 17%

So the poll here is about 3% more educated than the 2018 exit poll, which if reweighted would move the bottom line less than a point.

https://int.nyt.com/data/documenttools/nyt-siena-poll-methodology-june-2020/f6f533b4d07f4cbe/full.pdf


NYT/Siena is using a 22% white college - 50% white non college sample for Michigan (rest of the sample is miniorities/refusal to answer). And they're using a 20% white college - 38% white non college sample for Florida.


I trust NYT/Siena to conduct good polls because Nate Cohn is obsessed with data research and analysis.