A bird's eye view of the 2018 House landscape

With the midterms right around the corner (that's the election, to those of you who aren't staring down the barrel of an exam, or perhaps, grading some exams), it is worth a post explaining that, unfortunately, I am still right about where I was many months ago with respect to the House of Representatives. Here's the basic political science rundown on House elections in a midterm. They tend to run against the party of the president, barring something like 9/11, which helped the Republican Party pick up seats in 2002, or the wildly unpopular impeachment of Clinton, which helped the Democrats pick up seats in 1998. Other than that, the party of the president loses seats. In special circumstances, like a truly wretched economy, unpopular war, or something like that, you get a lot of seats changing hands. Mostly, though, there is a range of possibilities. Here is the basic problem. How many seats need to change hands for control of the chamber to flip? If the Democrats pick up 23 seats, they get a bare majority in the House. 218. Speaker Pelosi. How many different ways are there to do that, when each seat has a different probability of going Dem? That's a hard problem. We can run computer simulations for that, and calculate the number of times the Dems get the majority, if that's your thing and you are confident in your ability to assess each individual House seat's likelihood of going Dem, but that latter part is, itself, hard without House-level polling data, or at least final spending figures because, as I have written before here, the research on campaign spending does show that even if incumbents don't benefit that much from campaign spending, challengers do.

Forests or trees? Maybe look at the forest. Here's the alternative approach. Look at the historical circumstances in which large numbers of seats change hands, and small numbers of seats change hands, and the problem there is the divergence. The economy is remarkably strong. I've done enough of those posts that I won't desecrate the equine corpse in front of me yet again (that sounded less creepy in my head), but then we have Trump. The RCP average for his approval rating is 42.8%, which is up slightly, but still low. That divergence means we are in an unusual position. A strong economy should mean a popular president and a demotivated opposition party, but Trump motivates Democrats. An unpopular president should mean a demoralized and demotivated presidential party, but Trump motivates the Republican Party. He is negative partisanship personified, in the research tradition of Alan Abramowitz and Steven Webster, if such a tradition existed for a relatively new scholarly concept. However, the basic observational problem is that our two primary forces are working against each other. I find myself, then, much more uncertain about the 2018 House landscape than about a normal midterm because of the tension between the economy and Trump's approval rating.

The Senate, of course, is much easier. The Republicans win. The map puts up a bunch of seats the Democrats have to defend because they are the same seats they won in 2006 and 2012, and picking up even more from that set is way too hard. And they'll probably lose Heitkamp instead. With McCaskill in danger too, (and Menendez!) the GOP keeps the Senate. The House, though? Way too hard to predict.

The factors that go into the models this year? They are in tension with each other in an historically unusual way. Keep that in mind.