(Note from Harry: Follow me along live on Election Night here.)
The 2018 campaign (mostly) comes to an end today. If the polls and our forecasts are right, the Democrats and Republicans will each have something to be happy about.
Democrats are favored to take back the House, while Republicans are favored to maintain control of the Senate.
Our final House forecast has Democrats earning 227 seats to the Republicans 208. That's a net gain of 32 seats from the 195 they hold right now. Democrats only need a net gain of 23 to win the 218 seats necessary for a majority.
But as we have noted all along, our forecasts come with a margin of error. Specifically, our 95% confidence interval finds that Democrats could win as few as 207 seats (11 short of a majority) to 259, according to our latest estimate.
There are two things you should note about the range. The first is that it's wide because there are so many close races.
There are, for example, 32 races that we think will have a margin of 4 points or less. There are 97 races in which the margin of error (95% confidence interval) is wider than the margin by which one of the candidates is projected to win.
Of those 32 races that are within 4 points, 19 are forecasted to be won by the Democratic candidate. If only about half of these 19 go the other way, Republicans could conceivably maintain control.
The second thing you should note is that the margin of error is asymmetric. That is, the difference between our median estimate (227) and the bottom range of our margin of error (207) is only 20 seats, while the difference between our median and the top range of our margin of error (259) is 32 seats.
There are an astounding 61 races where the Republicans are favored, but where we think the Democratic candidate is within the margin of error of winning. If there is a small, but systematic, error in the polling, it's not inconceivable that Democrats could do far better than we have forecasted them to do.
The asymmetric range of our results is why if we were projecting our midpoint using an average instead of a median that would forecast a net gain of 34 seats for the Democrats.
Make no mistake though, the House race is still close enough that Republicans could win. The closeness of the result is such that the seven races within 4 points in California, Maine and Washington could be determinative. It may take some time to count the ballot for various reasons (e.g. mail-in ballots and ranked choice voting) in these contests. That means we may not know the winner for days.
Our final Senate forecast is something else altogether. It has Republicans controlling 52 seats and Democrats (and Independents who caucus with them) holding 48 seats in the next Congress. If this forecast were exactly right, it would mean that Republicans would have a net gain of a seat since the last Congress.
This overall forecast takes into account each state's forecast, the margin of error of each state's forecast and the correlation between the different state results. In cases where one side is expected to win many close victories, the overall forecast thinks there's a good chance they will lose a few of these contests.
That's exactly what's going on this year. Right now, our forecast has Democratic candidates winning by 2 points or less in Arizona, Florida, Indiana, Missouri and Nevada. No Republican is forecasted to win by less than 6 points. Based upon past history, the model expects that two of these close forecasted Democratic victories will turn into Democratic defeats. If all these states end up going to the Republicans, they will win 55 seats.
Indeed, we shouldn't underrate the possibility of Democrats winning the House and Republicans doing very well in the Senate tomorrow night.
If each of our state forecasts is right, however, Democrats will end up 50 seats to the Republicans 50 seats. Vice President Mike Pence would break the tie and give Republicans the barest of majorities in the Senate in this case.
Is it possible that Democrats win control of the Senate? Yes. Our forecasted margin of error gives Democrats the possibility of controlling up to 52 Senate seats in the next Congress.
Right now, the two most likely ways Democrats win the majority in Senate elections held tomorrow involve either Democrat Phil Bredesen winning in Tennessee or Democrat Beto O'Rourke winning in Texas. Both of those races are within 6 points, which means they are within the margin of error.
Perhaps the most intriguing scenario for election night is that if neither party wins a Senate majority tomorrow. There is little doubt at this point that the Mississippi Special Senate jungle primary will require a runoff in late November between the two top vote getters on Tuesday. In the jungle primary, all Democrats and Republicans run against each other. If no one receives a majority, there will be a runoff.
There are two Republicans (Chris McDaniel and Sen. Cindy Hyde-Smith) who will split the Republican vote in this very red state. The question is who advances to the runoff.
For most of the campaign, the polls pointed towards Hyde-Smith making the runoff against Democrat Mike Espy. In such a scenario, Republicans would be heavily favored to hold onto the seat.
There has been some late polling though that has Hyde-Smith and McDaniel neck-and-neck. If McDaniel beats out Hyde-Smith, this seat becomes very winnable for Democrats.
And keep in mind that if we project every other state correctly and it's a Espy-McDaniel runoff, said runoff will be for control of the Senate.
This story has been updated to reflect the most recent forecast figures.