IA-Selzer: Trump +7

[new_patomic]

So Ernst +4 in a race which is Trump +7

Hmm

[new_patomic]

At the very least, props to them for not herding their numbers.

[new_patomic]

They're the gold standard... until they're not.

That doesn't mean they're wrong here, or that the results won't end up in the margin of error of where they have it. But it isn't impossible for them to have misjudged the race, or for there to be healthy skepticism.

[new_patomic]

I would think that if Trump and Republicans were gaining with Independents and White voters in Iowa, it would show up in some other midwest polling.

But the best I can think of is like, the Republican internals showing Bustos only up 5-6 points, and even that doesn't seem right.

[new_patomic]

They've never weighed by education, I believe.

And wouldn't that being a problem mean Trump is up by even more.

[new_patomic]

Districts:

IA-01: R+15
IA-02: R+1
IA-03: D+6
IA-04: R+17

Hello, outlier. This is just not possible even remotely.

I JUST WARNED EVERYONE its a sample size of 200 for the district polls and it should be discounted, I even recall having this exact conversation a few months . There is obviously no way Iowa 2nd is 14 points to the left of the first but thats merely MOE and the district polls are merely crosstabs and should be seen as such, I have no idea why Selzer even reports them without at least doubling their total poll size. Even if the 1st is way too R friendly, for Trump +7 the 4th is way too D friendly. The 2nd seems about right and I am not sure about the 3rd which is the only Iowa district that has a major city so it could trend differently.

However Iowa 2nd and the 1st had the same Obama and Trump numbers although the 2nd has more red rurals while the 1st has less blue D counties.

It should be noted, that they say they weigh by Congressional District.

[new_patomic]

Look no one should be unskewing their polls



But in a case where you weigh by Congressional District, wouldn't that suggests that their CD results, however small of a sample, do impact the final statewide margin here? And so an unexpected result or weird sample in one district could actually present a problem for the poll?

[new_patomic]

You can always get weird subsamples in polls even while the overall sample is still legit.

But at this point I think it's fair to say that, given weighting by CD, their CD1 result does look like a genuine outlier. I don't think even the DCCC is that worried about Finkenauer, and she'd probably be triaged at this point if Trump were actually winning there by 15.

Add it to the average, move on.