People have been getting really confused about my tweet about Harvard-Harris and the Fox News poll for about 10 hours now, so I suspect some further explanation is in order. 1/
First, to be clear, I was talking about the job approval numbers and not the generic ballot numbers. I think the GB is an interesting heuristic, and is generally consistent with the national environment. I also don't find it to be terribly useful. 2/
Lots of smart people use it in their modeling, which is fine, as there are lots of good models out there, but I don't, mainly because in my experience aggregated models of congress don't do so well. I use an individual, district-level model. 3/
For my purposes, I find that a 1 percent increase in presidential job approval increases the odds of a candidate winning that district by about 1.25 percent. It's an important, but not overwhelming factor. 4/
My working assumption had been that Trump had a *lot* of downside to his job approval, and not much upside. IOW, I would have been much less surprised to have him at 30% on Election Day than, say, 45%. 5/
So my baseline expectation for Republicans in November of 2018 has been a bloodbath, something in the realm of around 40 seats, give-or-take. That would be the worst midterm election for Rs since 1974. 6/
But you have to be willing to update working assumptions. So when you see the two best non-Rasmussen job approval numbers since May of '17 back-to-back, you should cock your head to the side and say "huh." 7/
Does that mean that you toss out your assumption and say "Trump's headed to 50% approval?" Heck no. 8/
But you also have to build in that, given the recent spate of good economic news, the apparent gelling of the Republican Party from the tax cut bill, and a few other things, that the normal expectation is that job approval would rise. 9/
That's what we've seen. On average, he's at 40 percent right now, which still isn't good, and would still be catastrophic to Rs. But *if* (to clarify, this is a conjunction that introduces a *conditional*) 10/
again *if* we start to get more polls showing him in the mid-40s, I'd substantially revise my expectations about his ceiling, as well as how well Republicans were likely to do in the midterms. 11/
43 percent job approval is roughly the Mendoza line for Rs being in the game to hold the House. We haven't even had polls testing the upper bound above that number in a while. 12/
So, it's at least noteworthy to get polls up there. If we get a bunch more polls like Q-Pac or WaPo? Then you shrug and say "these were outliers, he probably still has a ceiling around 40 percent" and move on. 13/
And if we get polls showing him in the low 30s? Also noteworthy! But to me, not as surprising, since I don't have much of an expectation that he has a floor. 14/14
