Trump’s Approval Rating Stands Around 35 Percent in Three Key Midwest States

[Mr. Morden]

It should be noted that NBC News/Marist had Hillary winning PA by 12 points in mid-October.

As we know, Trump won by almost a point.

NBC didn't poll MI and WI in 2016.

Hillary was probably ahead by a decent amount in PA around mid-October. If your trying to tell us that Marist is inaccurate because of 2016, well then we wouldn't be able to trust most pollsters going forward, would we?

I'm not trying to say that Marist is necessarily inaccurate (they also had some good results, such as in AZ).

On the other hand, they also flopped badly in NC (predicting a 6-point Hillary win there), in a poll just a few days before election day (she lost by 3 or 4).

So it's possible that Trump's approval is actually at 40-42% in these 3 states right now ... who knows ?

I see no reason to believe any pollster is underestimating Trump's support, especially now since they are hyper aware of the polling failure that was 2016.

Depends on what's being done to fix it.  The AAPOR report claims that the big reason why the polls were so far off in the Rust Belt states this time was that the polling firms that did the most polls there didn't weight by education levels.  They weighted by other demographic factors, but not education levels, and since education was more strongly correlated with voting preference this time than normal, that was a problem:

https://uselectionatlas.org/FORUM/index.php?topic=263658.0

So is this poll weighting by education level?  Does the demographic breakdown on education level look reasonable?

[Mr. Morden]

2016 exit polls Trump fav/unfav.  WI 35/64.  PA 42/56. MI 39/59. This poll is all adults and not even LV.  It seems Trump gained a bit of ground since election night.

Exactly. Why is it so hard to understand that people will still vote for someone they dislike if the other person is seen as worse?

If you can name me a potential candidate with likability ratings similar to Hillary, then you have a point. But as of now, even polarizing Elizabeth Warren isnt anywhere near Hillary levels.

Well, the "as of now" thing is key.  The GOP hasn't yet gone to town on whoever the 2020 Dem. nominee is going to be in the way that they're going to ~30 months from now.  Once someone becomes the de facto nominee, then the GOP attack machine will focus its attention on that person, but that hasn't yet happened, because it's too early to know who that person will be.  Not saying that means they'll end up with favorability #s as bad as what Clinton had last year, but will they have worse favorable #s than "Generic Democrat", who is the only person "running" against Trump right now in most people's minds?  Probably.

There's also likely to be a "coming home" effect, whereby some Republican-leaning voters who currently feel free to hold a negative opinion of Trump end up excusing his sins once an actual flesh and blood Democratic challenger emerges.