Most professional election analysts (or at least the ones I know) expect the Democrats to take over the House. That’s not partisan hackery or an artifact of media bias; it’s an expectation based on a solid read of historical and current data. The president’s party almost always loses seats in midterm elections, and those losses are often big enough to take the majority from the president’s party (in two of the last three midterms, the president’s party lost more than 30 seats, which is more than the GOP’s roughly two-dozen seat edge). Moreover, Donald Trump doesn’t look especially popular by historical standards, so it’s not hard to imagine his party losing the lower chamber.
At the same time, most election analysts also believe that the probability of a Democratic takeover isn’t 100 percent. The country is pretty evenly divided on politics, so both sides have been able to compete in most recent national elections. FiveThirtyEight pegs the probability of a GOP win at 20 percent, and the Economist puts it at 29 percent.
Events with a 20 percent or 29 percent probability happen all the time, but for a lot of people they exist in a sort of gray probability zone. They know that the likelihood is above zero, but they’re not sure how to think about the probability. Moreover, most of the political world has thought longer and harder about what a Democratic House majority would look like in 2019 than they have about what a GOP House majority might look like.
Let’s look at two things: 1) how to get a handle on the actual likelihood of a low-to-medium probability event like a GOP House takeover and 2) what a GOP House win might look like if it were to materialize.
First, interpreting probabilities is a skill. If you’re not a data analyst or a professional poker player, you probably need some shorthand to help you get a handle on what different probabilities actually mean.
So here’s a quick guide: Each line is a probability and a day-to-day example of an event that occurs with that same probability
0 percent: total certainty that something won’t happen
3 percent: rolling snake eyes on two six-sided die
12.5 percent: flipping a coin three times and getting heads on all three flips
17 percent: rolling a six on a six-sided die
25 percent: flipping a coin twice and getting heads both times
33 percent: rolling a one or two on a six-sided die
38 percent: drawing a face card or a 10 off the top of a 52-card deck of playing cards
50 percent: the chance of flipping a fair coin once and getting heads
62 percent: drawing any card from two to nine off the top of a 52 card deck of playing cards
67 percent: rolling a six-sided die once and getting something between three and six
75 percent: flipping a coin twice and not getting two heads in a row
83 percent: rolling a six-sided die and getting anything between one and five
87.5 percent: flipping a coin three times and not getting heads three times in a row
97 percent: rolling two dice and not getting snake eyes
100 percent: total certainty
So a GOP House takeover is, according to these models, is somewhere relatively close to the odds of flipping a coin twice and getting heads both times. That’s far from a 0 percent chance, but it’s also noticeably different from the odds of winning a simple coin toss.
These numbers raise an obvious (and understudied, in my view) question: What would a GOP House win actually look like, from a polling point of view, on and after Election Day?
It’s possible that the GOP moves polls in their direction and manages to win the overall vote. But you probably shouldn’t bet on that: Polls usually move against the president’s party as Election Day gets closer.
A GOP victory would more likely reveal that the polls were systematically underestimating the GOP’s win probability. Earlier in the cycle, I estimated that if the Democrats won the House popular vote by seven points, they’d be even-money favorites to take the House. Right now, the Democrats are ahead by nine points. It’s possible to imagine scenarios where the vote was distributed in just the right way the GOP still managed to take the House while losing the popular vote by nine. But it’s easier to imagine a scenario where the polls happen to underestimate the Republicans overall and they hold the chamber while still losing the overall popular vote.
One important note here: It’s very hard to predict which way the polls will err. During the 2016 election, I read numerous takes about how the polls were underestimating Hillary Clinton. They underestimated Trump. In 2014, the polls were supposed to underestimate Democrats until they underestimated Republicans. And in 2012, the shoe was on the other foot—the polls were biased in favor of Mitt Romney in key states, not against him as some alleged.
But no matter where the polls end up (whether it’s a miss in the Democratic direction, the GOP direction or an exact hit), there’s probably going to be some serious post-election political chaos.
If Democrats win, they’ll have to have some tough internal debates about how to deal with President Trump—if impeachment is actually on the table, how aggressively they want to use procedure to block Trump’s policy priorities, what the right balance of cooperation and aggression is for each member’s district, etc.
If Republicans win, they may do so with a relatively thin margin. That’ll encourage defections on key votes and it’ll up the ante on retirements, resignations and special elections. Simply put, there’d be chaos. Republicans would be just a handful of defections away from losing every controversial vote. Each retirement, resignation or special election would have the potential to meaningfully shift the balance of power within the chamber.
It’s also possible that Congress will find some way to behave well and rest at some sort of an equilibrium. But if 2016 taught us anything, it’s that we shouldn’t underestimate the possibility of chaos reigning in U.S. politics.
