IA-Selzer: Trump +7

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

This is a really really bad poll for Dems, perhaps the worst one all year. I wasn't really expecting Biden to win Iowa and would have been pleasantly surprised if he did (just did my prediction map with it going R yesterday), but the really bad thing about this is it suggests that Trump's gains (and over the longer term the GOP's gains) among northern/midwestern rural WWC whites were not a one time wonder. This means we should be worried in particular about Wisconsin and Michigan as well, and to a lesser extent Minnesota and Pennsylvania (though in those states I think the Minneapolis/Philadelphia suburbs will be enough of a countertrend regardless). Also despite most polls of ME-02 having Trump up, this suggests he could win that again. Yes, those other state are different from Iowa in various ways, but it is unlikely that WWC voters will swing/trend in a massively different direction in those different states.

This is just a single poll, but it is a very important one because it is the first significant hint that Trump may have a chance in quite a while.

Overall, this greatly raises Trump's chances of winning from essentially non-existent to give Trump a non-trivial chance of pulling out a skin-of-his-teeth win. And that is too high of a chance.

If this does come to pass, over the longer term, this points to the electoral college and the Senate becoming more systemically biased in favor of the GOP.

If nothing else good comes out of this, hopefully one thing this will accomplish is to persuade more Dems of the necessity of abolishing the electoral college, and abolishing the Senate and/or adding lots of new states and/or changing it so that states don't all have exactly 2 Senators.

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

This is a really really bad poll for Dems, perhaps the worst one all year. I wasn't really expecting Biden to win Iowa and would have been pleasantly surprised if he did (just did my prediction map with it going R yesterday), but the really bad thing about this is it suggests that Trump's gains (and over the longer term the GOP's gains) among northern/midwestern rural WWC whites were not a one time wonder. This means we should be worried in particular about Wisconsin and Michigan as well, and to a lesser extent Minnesota and Pennsylvania (though in those states I think the Minneapolis/Philadelphia suburbs will be enough of a countertrend regardless). Also despite most polls of ME-02 having Trump up, this suggests he could win that again. Yes, those other state are different from Iowa in various ways, but it is unlikely that WWC voters will swing/trend in a massively different direction in those different states.

This is just a single poll, but it is a very important one because it is the first significant hint that Trump may have a chance in quite a while.

Overall, this greatly raises Trump's chances of winning from essentially non-existent to give Trump a non-trivial chance of pulling out a skin-of-his-teeth win. And that is too high of a chance.

If this does come to pass, over the longer term, this points to the electoral college and the Senate becoming more systemically biased in favor of the GOP.

If nothing else good comes out of this, hopefully one thing this will accomplish is to persuade more Dems of the necessity of abolishing the electoral college, and abolishing the Senate and/or adding lots of new states and/or changing it so that states don't all have exactly 2 Senators.

Maybe it means the WWC voters are doing one last “swan song” for Biden like Dixiecrats did for Carter and suburbanites for Romney.

Yeah, I think Biden will probably nonetheless (despite this 1 poll) do relatively well among Midwestern WWC voters. And, for example, Biden may do significantly better in WI as compared to IA this time around due to the COVID outbreak in WI. But over the longer term it definitely suggests that those voters are likely to keep trending Republican. The new electoral map in the future is going to be a contest between Dems trying to win with a map based on winning competitive sunbelt states like TX, GA, NC, and FL on the one hand and Republicans trying to base their route to 270 on winning states like WI, MI, MN, ME, and PA while holding on to a few of those sunbelt states.

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

The other thing this suggests to me is that if Dems do nonetheless win a trifecta in 2020, after taking care of the immediate/necessary COVID relief legislation etc, when they eventually start on policy legislation the first order of business should probably be immigration reform. Winning the sunbelt in the future is going to be very important for Dems in future elections, and so it is time to try and lock in more Hispanic support, and get more people on the pathway to citizenship and eventually voting in those sunbelt states (most importantly, Texas).

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

I said almost two years ago that Finkenauer would be one of the most likely Dem flips.

Firstly, she's not a good candidate. Also there are no major cities in her district (Cedar Rapids doesn't count), not many suburbs, virtually no minorities and it's not particularly well educated. A district like that is not staying Dem in 2020 outside of New England.

But Trump only won here by about 4 points. This district is just like Cheri Bustos’ and Ron Kind’s districts, and neither of them seem in much trouble.

They may not be in trouble now, but if it is a Biden midterm in 2022, Bustos/Kind will be in trouble then. If IL Dems have half a brain, they will draw a ridiculous tentacle from Bustos' district into Chicagoland.

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

They may not be in trouble now, but if it is a Biden midterm in 2022, Bustos/Kind will be in trouble then. If IL Dems have half a brain, they will draw a ridiculous tentacle from Bustos' district into Chicagoland.

There is little need to draw it to Chicago Land, just strip the rurals and add Mcclean county. There isn't even enough room in Chicago land.

If you are trying to protect Bustos against a possible midterm red wave, the last thing you want to do is "shore her up" by giving her a bunch of college students in Bloomington who won't bother to vote in a midterm (just like in 2010 and 2014). No, no, no. The point is to make her less exposed to midterm Dem turnout drop-off, not more.

Instead, what you do is remove Peoria from her seat altogether and make her seat a Moline-Rockford district with a tentacle from Rockford into Chicagoland. There are various ways it could go, personally I like drawing it along the Wisconsin border from Rockford into Waukegan/North Chicago to add some non-white voters into the district. That makes it a lot less dependent on fickle white voters. But alternatively you can draw it into places like DeKalb, Aurora, and/or Elgin.

Meanwhile you put Bloomington and Peoria into an entirely different district. Either you put them into IL-13 and make that district actually be Dem enough that a Dem can win it, or else (probably wiser) you draw those into Chicagoland as well with a separate set of ridiculous tentacles. A good candidate district for that job is IL-14, because it has way too much dangerous exurban/semi-rural territory in the first place. But frankly it is probably better to have 2 districts split that job, both with their own sets of ridiculous tentacles.

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

Another laughable cross-tab: Trump is winning the youth vote!

That probably is wrong, but the general idea of Trump/Republicans doing relatively well among white midwestern youngs is less implausible than it sounds. It is normal that if a state is R-trending you can see Republicans doing better (or at least not as bad as you would think) with young voters. You can see a similar thing going back many years in other states that were at one point safe Dem or competitive as they have trended R such as West Virginia, Arkansas. This occurs because there are still some older Demosaurs in those states, whereas the young whites are more likely to be open to the GOP than the Demosaurs.

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

I just checked and WTF is this.

One GA poll with Biden at +5 or +7... "Outlier, two-page thread"

One Iowa poll with Trump at +7... *17 PAGES!*

People. Get...A...Grip.

The reason this poll is getting so much attention is not because of the poll itself or its direct results.

The reason it is getting a lot of attention is because it is the only poll from a credible pollster we have seen in a long time that suggests Trump has any real path to winning at all. There is a big difference between "basically no chance of winning" and "small but real" chance of winning.

Another thing that some people noted is that Selzer didn't weight for education, and that may play a part in this result. Indeed, that is quite plausible. She may have gotten a less educated sample.

One of Selzer's methodological calling cards has long been that she deliberately tries not to assume too much about the electorate and especially not to limit who she includes in a poll or counts as a "likely voter" too much. That can have its pluses and its minuses.

But one of its pluses is that it is basically means her poll is listening "close to the ground." It lets her pick up on late breaking trends that other polls might miss, at the cost of possibly being over-sensitive.

The one thing that we have all known for a long time that Trump needs to have any realistic chance of winning is big enthusiasm and big election day turnout from WWC voters.

So, suppose that this poll is  as Trumpy as it is because it has "too many" non-college white voters in the sample. What might that mean? Well, it might mean that non-college white voters in Iowa are pretty enthusiastic about voting and thus were more likely to respond to the poll.

That could be a harbinger of strong WWC turnout on election day. It isn't necessarily, but it could be. And if there were a bunch of late-breaking enthusiasm among non-college white voters and if there were about to be a big surge of WWC election day turnout, one of the first places we would expect to see hints of that would be in a poll like the Selzer poll that deliberately tries not to limit its likely voter pool too much or make too many assumptions about it. And if there were a surge of WWC election day voting, it probably would not occur just in Iowa but also would occur to some degree in other states. It may be easier to detect in Iowa since Iowa is more rural, whiter, and doesn't have large metro areas, however.

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

And also IA's correlation to the MIWIPA trio is overrated as hell.

Iowa voted 9+ points to the right of all three of those states in 2016, and also had a much weaker showing for Democrats in 2018 than the MIWIPA trio in which the only close statewide race was the WI Governor race which the Democrats won.

True, the correlation to those states as a whole is fairly weak.

However, the correlation to rural parts of those states is a lot stronger.

The reason why the correlation is weak is because those other states have large metros with lots of educated whites and/or minorities which Iowa doesn't have such as Madison, Milwaukee, Detroit, Philadelphia, and Pittsburgh.

This poll is not an indicator that Trump will necessarily win or do particularly well in PA, MI, or even WI. But it is an indicator that he could do fairly well in rural counties within those states; possibly either they don't swing back to Biden, or at least not by as much as one would expect based purely off of other datapoints. And that is also a longer term indicator that in future elections Republicans may do well in those rural white areas also, and eventually make additional gains above what Trump already got.

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

yaaaasssss let's make it 25 pages!

This thread should be longer. If you agree, empty quote this.

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

That Selzer was right and so many others were wrong may suggest education weighting is not the panacea so many thought. Selzer doesn't do it.

Something I posted a few weeks back:

We've been over this.

The problem with 2016 was that pollsters were not controlling for education because historically, education level was not a reliable predictor of voting patterns. That has changed, and now the more educated you are the more likely you are to vote Democrat.

That gap is especially true among white voters.

Pollsters are now controlling for education.

So unless there's something else they're missing, they're probably not overestimating Biden's white support.

Here's the thing: weighting by education only solves your problem if the small sample of non-college voters you do successfully get is conditionally representative of the non-responsive non-college population.

As I doubt pollsters' ability to reach non-college whites has increased since 2016, this means inflating the importance of the few non-college respondents you do have, which is a group which tends to be disproportionately elderly. If, for example, elderly non-college voters have moved left since 2016 while others have not (or even moved right), education weighting may actually make polls less accurate than without education weighting.

This seems spot on. It seems that polls in general currently have a significant problem with non-response bias. The problem is that the overall response rate to polls is so low (and continuously going lower) that even a fairly small response bias can significantly bias the results.

Polls should be treated with a lot more general skepticism for the time being unless/until pollsters can come up with an alternative way of reaching people that can attain a higher general response rate while also achieving a truly randomized and representative sample.

Methods like post-stratification are very nice to paper things over and try to get at least something filling in the response gaps, but it is still very easy for small numbers of respondents from a demographic with a low response rate to not be representative of the demographic as a whole, which is precisely the conditional unrepresentativeness problem that you rightly highlight.