Election day 2018 is finally here– a mildly important election that has been sold as the most important election ever. With all the hype, there can be a lot of information to process. You can have nitwits on the left saying Beto O’Rourke is going to win (don’t bet on it) and dimwits on the right telling you a #redwave is about to happen (don’t bet on that either).
Here is a quick primer on what to expect. Nate Silver’s Fivethirtyeight “classic” election model (i.e. polls plus “fundamentals”), which is based on thousands simulations, expects Democrats to pick up 39 House seats (a 33 seat Democratic majority) on average with 39 or 40 House seats (a 33 or 35 seat Democratic majority) being the two most likely outcome, but 80% of the outcomes occur between Democrats picking up 21 House seats (a 3 seat Republican majority) and 59 House seats (a 73 seat Democratic majority). This model gives Democrats a 7 in 8– or 88% chance– of winning the House of Representatives but, Republicans still have a path with a 12% chance.
Given that massive confidence interval, who controls the House of Representatives is clearly up for grabs, but Democrats will– more likely than not– control it next Congress. As Nate Silver said, for Republicans to hold the House they will need a “systematic polling error” — i.e. “…polls to be off everywhere, or at least in certain key clusters of states…”. The good news for Republicans is that the polling error does not need to be that big– only two or three points in their direction– which is well within historic standards.
If you look at the “lite” version (i.e. polls only, no “fundamentals”) of the model instead of the “classic” version, Democrats should pick up 38 House seats (a 31 seat Democratic majority) on average with 33 House seats (a 21 seat Democratic majority) being the most likely single outcome. The 80% confidence interval is from Democrats picking up 61 House seats (a 77 seat Republican majority) to Democrats picking up 17 House seats (a 11 seat Republican majority). In the lite model, Democratic chances take a fairly significant blow and fall to 4 in 5– or an 80% chance– while Republicans rise to an 20% chance.
But the Senate model tells a very different story. The classic model shows Republicans picking up 0.5 Senate seats on average with no net change as the single most likely outcome. Although the 80% confidence interval goes from a Democratic pick up of 2 seats (for a 1 seat majority) to a Republican gain of 4 seats (to extend their majority to 10 seats), the 4 most likely outcomes are no net gain for Republicans or better and 6 of the 7 outcomes in the interval give Republicans the majority. Republicans hold the chamber in 81% of outcomes while Democrats take it in 19% of outcomes.
Unlike in the House, it is much clearer who will hold the chamber. Even if Democrats win every “lean D” and “toss-up” race while also not coughing up any “likely D” or “solid D” races, they will still have only 50 Senate Seats and will need to hold the North Dakota seat– which Republicans have an almost 75% chance to win– or pull an upset– with Mississippi being the most likely candidate.
Flipping to the lite model, it shows Republicans picking up 0.7 Senate seats on average with no net change as the single most likely outcome. Otherwise, the models are similar to the point explaining their differences would be redundant. The most significant difference is the lite model moves North Dakota from “lean R” to “likely R”, Nebraska from “solid R” to “likely R” and Montana from “likely D” to “lean D”. Here, the odds are more or less than same, but Republican odds fall to 79% while Democratic odds rise to 21%.
Another important thing to know is mistakes to avoid. Here are a few:
- House control is not the generic ballot. If Republicans keep control, it does not mean polls were wrong. To judge the generic ballot, you need to compare the “House popular vote” to an average of generic ballots. And when evaluating polls, remember that the margin of error is a thing.
- Republicans will keep the House and keep the Senate are 15% each– that means, if we treat each as independent, the odds of Republicans keeping both are between getting four and five straight heads on a coin flip. Hardly impossible. Don’t round 85% up to 100%. To quote Silver, “an 86 percent chance might seem like a sure thing, but it isn’t — would you board a plane that had a 14 percent chance of crashing?”
- Uncertainty goes both ways. Any uncertainty that can benefit one side can benefit the other.
All that said, I also think it is important to have a framework for interpreting election results beforehand. In the House, I think keeping their losses to less than 20 is a good night while between 20 and 22 lost seats would be a decent night. On the flip side, I think 23 to 30 gains for the Democrats would be a decent night for them while anything more than 30 would be a good night.
On the flip side, I think a good night for Senate Republicans is picking up 3 or more seats while only picking up 1 or 2 would be a decent night. No net change would be a neutral night– although I am tempted to call it a good night for Democrats– and any net gain for Democrats will mean a great night for them.
In the end of the day, I think whoever wins Missouri will win Florida, Arizona and probably Nevada. I think Republicans could come out with an extra seat but I am inclined to say we will see a 51-49 Senate next Congress. In the House, I think the Democrats will pick up 30 to 40 seats and it is more likely they roll up to 50 than underperform with 20.
Lastly, I’ll leave you with an exit quote from Silver: “The range of possible outcomes is wide… [but] nonetheless, the considerable majority of the range is in Democratic territory.”